ss2v110: bin/SeqSero2_package.py annotate

annotate bin/SeqSero2_package.py @ 14:047956dacae4 draft default tip

planemo upload commit 70dc513aa7d7ac6785847dfd86323687613b6b68-dirty

author	cstrittmatter
date	Thu, 21 May 2020 22:01:02 -0400
parents	08832c0d3cbd
children

rev	line source
0 fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1 #!/usr/bin/env python3
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	2
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	3 import sys
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	4 import time
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	5 import random
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	6 import os
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	7 import subprocess
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	8 import gzip
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	9 import io
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	10 import pickle
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	11 import argparse
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	12 import itertools
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	13 from distutils.version import LooseVersion
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	14 from distutils.spawn import find_executable
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	15 sys.path.insert(1,sys.path[0]+'/..')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	16
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	17 try:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	18 from .version import SeqSero2_version
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	19 except Exception: #ImportError
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	20 from version import SeqSero2_version
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	21
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	22 ### SeqSero Kmer
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	23 def parse_args():
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	24 "Parse the input arguments, use '-h' for help."
8 357e38526e2a planemo upload commit c50df40caef2fb97c178d6890961e0e527992324-dirty cstrittmatter parents: 7 diff changeset	25 parser = argparse.ArgumentParser(usage='SeqSero2_package.py -t <data_type> -m <mode> -i <input_data> [-d <output_directory>] [-p <number of threads>] [-b <BWA_algorithm>]\n\nDevelopper: Shaokang Zhang (zskzsk@uga.edu), Hendrik C Den-Bakker (Hendrik.DenBakker@uga.edu) and Xiangyu Deng (xdeng@uga.edu)\n\nContact email:seqsero@gmail.com\n\nVersion: v1.1.1')#add "-m <data_type>" in future
0 fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	26 parser.add_argument("-i",nargs="+",help="<string>: path/to/input_data",type=os.path.abspath) ### add 'type=os.path.abspath' to generate absolute path of input data.
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	27 parser.add_argument("-t",choices=['1','2','3','4','5','6'],help="<int>: '1' for interleaved paired-end reads, '2' for separated paired-end reads, '3' for single reads, '4' for genome assembly, '5' for nanopore fasta, '6' for nanopore fastq")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	28 parser.add_argument("-b",choices=['sam','mem'],default="mem",help="<string>: algorithms for bwa mapping for allele mode; 'mem' for mem, 'sam' for samse/sampe; default=mem; optional; for now we only optimized for default 'mem' mode")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	29 parser.add_argument("-p",default="1",help="<int>: number of threads for allele mode, if p >4, only 4 threads will be used for assembly since the amount of extracted reads is small, default=1")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	30 parser.add_argument("-m",choices=['k','a'],default="a",help="<string>: which workflow to apply, 'a'(raw reads allele micro-assembly), 'k'(raw reads and genome assembly k-mer), default=a")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	31 parser.add_argument("-n",help="<string>: optional, to specify a sample name in the report output")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	32 parser.add_argument("-d",help="<string>: optional, to specify an output directory name, if not set, the output directory would be 'SeqSero_result_'+time stamp+one random number")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	33 parser.add_argument("-c",action="store_true",help="<flag>: if '-c' was flagged, SeqSero2 will only output serotype prediction without the directory containing log files")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	34 parser.add_argument("-s",action="store_true",help="<flag>: if '-s' was flagged, SeqSero2 will not output header in SeqSero_result.tsv")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	35 parser.add_argument("--check",action="store_true",help="<flag>: use '--check' flag to check the required dependencies")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	36 parser.add_argument('-v', '--version', action='version', version='%(prog)s ' + SeqSero2_version)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	37 return parser.parse_args()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	38
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	39 ### check paths of dependencies
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	40 check_dependencies = parse_args().check
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	41 dependencies = ['bwa','samtools','blastn','fastq-dump','spades.py','bedtools','SalmID.py']
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	42 if check_dependencies:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	43 for item in dependencies:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	44 ext_path = find_executable(item)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	45 if ext_path is not None:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	46 print ("Using "+item+" - "+ext_path)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	47 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	48 print ("ERROR: can not find "+item+" in PATH")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	49 sys.exit()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	50 ### end of --check
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	51
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	52 def reverse_complement(sequence):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	53 complement = {
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	54 'A': 'T',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	55 'C': 'G',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	56 'G': 'C',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	57 'T': 'A',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	58 'N': 'N',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	59 'M': 'K',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	60 'R': 'Y',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	61 'W': 'W',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	62 'S': 'S',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	63 'Y': 'R',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	64 'K': 'M',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	65 'V': 'B',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	66 'H': 'D',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	67 'D': 'H',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	68 'B': 'V'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	69 }
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	70 return "".join(complement[base] for base in reversed(sequence))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	71
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	72
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	73 def createKmerDict_reads(list_of_strings, kmer):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	74 kmer_table = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	75 for string in list_of_strings:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	76 sequence = string.strip('\n')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	77 for i in range(len(sequence) - kmer + 1):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	78 new_mer = sequence[i:i + kmer].upper()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	79 new_mer_rc = reverse_complement(new_mer)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	80 if new_mer in kmer_table:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	81 kmer_table[new_mer.upper()] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	82 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	83 kmer_table[new_mer.upper()] = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	84 if new_mer_rc in kmer_table:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	85 kmer_table[new_mer_rc.upper()] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	86 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	87 kmer_table[new_mer_rc.upper()] = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	88 return kmer_table
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	89
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	90
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	91 def multifasta_dict(multifasta):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	92 multifasta_list = [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	93 line.strip() for line in open(multifasta, 'r') if len(line.strip()) > 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	94 ]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	95 headers = [i for i in multifasta_list if i[0] == '>']
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	96 multifasta_dict = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	97 for h in headers:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	98 start = multifasta_list.index(h)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	99 for element in multifasta_list[start + 1:]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	100 if element[0] == '>':
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	101 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	102 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	103 if h[1:] in multifasta_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	104 multifasta_dict[h[1:]] += element
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	105 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	106 multifasta_dict[h[1:]] = element
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	107 return multifasta_dict
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	108
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	109
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	110 def multifasta_single_string(multifasta):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	111 multifasta_list = [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	112 line.strip() for line in open(multifasta, 'r')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	113 if (len(line.strip()) > 0) and (line.strip()[0] != '>')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	114 ]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	115 return ''.join(multifasta_list)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	116
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	117
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	118 def chunk_a_long_sequence(long_sequence, chunk_size=60):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	119 chunk_list = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	120 steps = len(long_sequence) // 60 #how many chunks
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	121 for i in range(steps):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	122 chunk_list.append(long_sequence[i * chunk_size:(i + 1) * chunk_size])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	123 chunk_list.append(long_sequence[steps * chunk_size:len(long_sequence)])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	124 return chunk_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	125
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	126
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	127 def target_multifasta_kmerizer(multifasta, k, kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	128 forward_length = 300 #if find the target, put forward 300 bases
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	129 reverse_length = 2200 #if find the target, put backward 2200 bases
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	130 chunk_size = 60 #it will firstly chunk the single long sequence to multiple smaller sequences, it controls the size of those smaller sequences
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	131 target_mers = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	132 long_single_string = multifasta_single_string(multifasta)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	133 multifasta_list = chunk_a_long_sequence(long_single_string, chunk_size)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	134 unit_length = len(multifasta_list[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	135 forward_lines = int(forward_length / unit_length) + 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	136 reverse_lines = int(forward_length / unit_length) + 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	137 start_num = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	138 end_num = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	139 for i in range(len(multifasta_list)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	140 if i not in range(start_num, end_num): #avoid computational repetition
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	141 line = multifasta_list[i]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	142 start = int((len(line) - k) // 2)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	143 s1 = line[start:k + start]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	144 if s1 in kmerDict: #detect it is a potential read or not (use the middle part)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	145 if i - forward_lines >= 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	146 start_num = i - forward_lines
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	147 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	148 start_num = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	149 if i + reverse_lines <= len(multifasta_list) - 1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	150 end_num = i + reverse_lines
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	151 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	152 end_num = len(multifasta_list) - 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	153 target_list = [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	154 x.strip() for x in multifasta_list[start_num:end_num]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	155 ]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	156 target_line = "".join(target_list)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	157 target_mers += [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	158 k1 for k1 in createKmerDict_reads([str(target_line)], k)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	159 ] ##changed k to k1, just want to avoid the mixes of this "k" (kmer) to the "k" above (kmer length)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	160 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	161 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	162 return set(target_mers)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	163
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	164
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	165 def target_read_kmerizer(file, k, kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	166 i = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	167 n_reads = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	168 total_coverage = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	169 target_mers = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	170 if file.endswith(".gz"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	171 file_content = io.BufferedReader(gzip.open(file))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	172 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	173 file_content = open(file, "r").readlines()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	174 for line in file_content:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	175 start = int((len(line) - k) // 2)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	176 if i % 4 == 2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	177 if file.endswith(".gz"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	178 s1 = line[start:k + start].decode()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	179 line = line.decode()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	180 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	181 s1 = line[start:k + start]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	182 if s1 in kmerDict: #detect it is a potential read or not (use the middle part)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	183 n_reads += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	184 total_coverage += len(line)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	185 target_mers += [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	186 k1 for k1 in createKmerDict_reads([str(line)], k)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	187 ] #changed k to k1, just want to avoid the mixes of this "k" (kmer) to the "k" above (kmer length)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	188 i += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	189 if total_coverage >= 4000000:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	190 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	191 return set(target_mers)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	192
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	193
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	194 def minion_fasta_kmerizer(file, k, kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	195 i = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	196 n_reads = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	197 total_coverage = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	198 target_mers = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	199 for line in open(file):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	200 if i % 2 == 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	201 for kmer, rc_kmer in kmers(line.strip().upper(), k):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	202 if (kmer in kmerDict) or (rc_kmer in kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	203 if kmer in target_mers:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	204 target_mers[kmer] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	205 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	206 target_mers[kmer] = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	207 if rc_kmer in target_mers:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	208 target_mers[rc_kmer] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	209 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	210 target_mers[rc_kmer] = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	211 i += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	212 return set([h for h in target_mers])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	213
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	214
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	215 def minion_fastq_kmerizer(file, k, kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	216 i = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	217 n_reads = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	218 total_coverage = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	219 target_mers = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	220 for line in open(file):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	221 if i % 4 == 2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	222 for kmer, rc_kmer in kmers(line.strip().upper(), k):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	223 if (kmer in kmerDict) or (rc_kmer in kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	224 if kmer in target_mers:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	225 target_mers[kmer] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	226 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	227 target_mers[kmer] = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	228 if rc_kmer in target_mers:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	229 target_mers[rc_kmer] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	230 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	231 target_mers[rc_kmer] = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	232 i += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	233 return set([h for h in target_mers])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	234
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	235
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	236 def multifasta_single_string2(multifasta):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	237 single_string = ''
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	238 with open(multifasta, 'r') as f:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	239 for line in f:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	240 if line.strip()[0] == '>':
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	241 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	242 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	243 single_string += line.strip()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	244 return single_string
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	245
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	246
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	247 def kmers(seq, k):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	248 rev_comp = reverse_complement(seq)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	249 for start in range(1, len(seq) - k + 1):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	250 yield seq[start:start + k], rev_comp[-(start + k):-start]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	251
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	252
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	253 def multifasta_to_kmers_dict(multifasta,k_size):#used to create database kmer set
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	254 multi_seq_dict = multifasta_dict(multifasta)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	255 lib_dict = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	256 for h in multi_seq_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	257 lib_dict[h] = set(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	258 [k for k in createKmerDict_reads([multi_seq_dict[h]], k_size)])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	259 return lib_dict
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	260
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	261
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	262 def Combine(b, c):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	263 fliC_combinations = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	264 fliC_combinations.append(",".join(c))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	265 temp_combinations = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	266 for i in range(len(b)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	267 for x in itertools.combinations(b, i + 1):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	268 temp_combinations.append(",".join(x))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	269 for x in temp_combinations:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	270 temp = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	271 for y in c:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	272 temp.append(y)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	273 temp.append(x)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	274 temp = ",".join(temp)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	275 temp = temp.split(",")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	276 temp.sort()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	277 temp = ",".join(temp)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	278 fliC_combinations.append(temp)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	279 return fliC_combinations
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	280
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	281
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	282 def seqsero_from_formula_to_serotypes(Otype, fliC, fljB, special_gene_list,subspecies):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	283 #like test_output_06012017.txt
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	284 #can add more varialbles like sdf-type, sub-species-type in future (we can conclude it into a special-gene-list)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	285 from Initial_Conditions import phase1,phase2,phaseO,sero,subs,remove_list,rename_dict
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	286 rename_dict_not_anymore=[rename_dict[x] for x in rename_dict]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	287 rename_dict_all=rename_dict_not_anymore+list(rename_dict) #used for decide whether to
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	288 seronames = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	289 seronames_none_subspecies=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	290 for i in range(len(phase1)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	291 fliC_combine = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	292 fljB_combine = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	293 if phaseO[i] == Otype: # no VII in KW, but it's there
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	294 ### for fliC, detect every possible combinations to avoid the effect of "["
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	295 if phase1[i].count("[") == 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	296 fliC_combine.append(phase1[i])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	297 elif phase1[i].count("[") >= 1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	298 c = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	299 b = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	300 if phase1[i][0] == "[" and phase1[i][-1] == "]" and phase1[i].count(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	301 "[") == 1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	302 content = phase1[i].replace("[", "").replace("]", "")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	303 fliC_combine.append(content)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	304 fliC_combine.append("-")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	305 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	306 for x in phase1[i].split(","):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	307 if "[" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	308 b.append(x.replace("[", "").replace("]", ""))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	309 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	310 c.append(x)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	311 fliC_combine = Combine(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	312 b, c
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	313 ) #Combine will offer every possible combinations of the formula, like f,[g],t: f,t f,g,t
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	314 ### end of fliC "[" detect
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	315 ### for fljB, detect every possible combinations to avoid the effect of "["
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	316 if phase2[i].count("[") == 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	317 fljB_combine.append(phase2[i])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	318 elif phase2[i].count("[") >= 1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	319 d = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	320 e = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	321 if phase2[i][0] == "[" and phase2[i][-1] == "]" and phase2[i].count(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	322 "[") == 1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	323 content = phase2[i].replace("[", "").replace("]", "")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	324 fljB_combine.append(content)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	325 fljB_combine.append("-")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	326 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	327 for x in phase2[i].split(","):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	328 if "[" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	329 d.append(x.replace("[", "").replace("]", ""))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	330 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	331 e.append(x)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	332 fljB_combine = Combine(d, e)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	333 ### end of fljB "[" detect
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	334 new_fliC = fliC.split(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	335 ","
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	336 ) #because some antigen like r,[i] not follow alphabetical order, so use this one to judge and can avoid missings
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	337 new_fliC.sort()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	338 new_fliC = ",".join(new_fliC)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	339 new_fljB = fljB.split(",")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	340 new_fljB.sort()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	341 new_fljB = ",".join(new_fljB)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	342 if (new_fliC in fliC_combine
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	343 or fliC in fliC_combine) and (new_fljB in fljB_combine
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	344 or fljB in fljB_combine):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	345 ######start, remove_list,rename_dict, added on 11/11/2018
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	346 if sero[i] not in remove_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	347 temp_sero=sero[i]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	348 if temp_sero in rename_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	349 temp_sero=rename_dict[temp_sero] #rename if in the rename list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	350 if temp_sero not in seronames:#the new sero may already included, if yes, then not consider
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	351 if subs[i] == subspecies:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	352 seronames.append(temp_sero)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	353 seronames_none_subspecies.append(temp_sero)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	354 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	355 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	356 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	357 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	358 ######end, added on 11/11/2018
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	359 #analyze seronames
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	360 subspecies_pointer=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	361 if len(seronames) == 0 and len(seronames_none_subspecies)!=0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	362 # ed_SL_12182019: modified to fix the subspecies output problem
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	363 #seronames=seronames_none_subspecies
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	364 seronames=["N/A"]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	365 #subspecies_pointer="1"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	366 subspecies_pointer="0"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	367 if len(seronames) == 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	368 seronames = [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	369 "N/A (The predicted antigenic profile does not exist in the White-Kauffmann-Le Minor scheme)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	370 ]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	371 star = ""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	372 star_line = ""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	373 if len(seronames) > 1: #there are two possible predictions for serotypes
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	374 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	375 #changed 04072019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	376 #star_line = "The predicted serotypes share the same general formula:\t" + Otype + ":" + fliC + ":" + fljB + "\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	377 if subspecies_pointer=="1" and len(seronames_none_subspecies)!=0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	378 star="*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	379 star_line="The predicted O and H antigens correspond to serotype '"+(" or ").join(seronames)+"' in the Kauffmann-White scheme. The predicted subspecies by SalmID (github.com/hcdenbakker/SalmID) may not be consistent with subspecies designation in the Kauffmann-White scheme. " + star_line
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	380 #star_line="The formula with this subspieces prediction can't get a serotype in KW manual, and the serotyping prediction was made without considering it."+star_line
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	381 if Otype=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	382 Otype="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	383 predict_form = Otype + ":" + fliC + ":" + fljB
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	384 predict_sero = (" or ").join(seronames)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	385 ###special test for Enteritidis
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	386 if predict_form == "9:g,m:-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	387 sdf = "-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	388 for x in special_gene_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	389 if x.startswith("sdf"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	390 sdf = "+"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	391 #star_line="Detected sdf gene, a marker to differentiate Gallinarum and Enteritidis"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	392 star_line="sdf gene detected. "
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	393 #predict_form = predict_form + " Sdf prediction:" + sdf
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	394 predict_form = predict_form #changed 04072019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	395 if sdf == "-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	396 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	397 #star_line="Didn't detected sdf gene, a marker to differentiate Gallinarum and Enteritidis"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	398 star_line="sdf gene not detected. "
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	399 #changed in 04072019, for new output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	400 #star_line = "Additional characterization is necessary to assign a serotype to this strain. Commonly circulating strains of serotype Enteritidis are sdf+, although sdf- strains of serotype Enteritidis are known to exist. Serotype Gallinarum is typically sdf- but should be quite rare. Sdf- strains of serotype Enteritidis and serotype Gallinarum can be differentiated by phenotypic profile or genetic criteria.\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	401 #predict_sero = "Gallinarum/Enteritidis" #04132019, for new output requirement
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	402 predict_sero = "Gallinarum or Enteritidis"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	403 ###end of special test for Enteritidis
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	404 elif predict_form == "4:i:-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	405 predict_sero = "I 4,[5],12:i:-" # change serotype name
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	406 elif predict_form == "4:r:-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	407 predict_sero = "N/A (4:r:-)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	408 elif predict_form == "4:b:-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	409 predict_sero = "N/A (4:b:-)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	410 #elif predict_form == "8:e,h:1,2": #removed after official merge of newport and bardo
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	411 #predict_sero = "Newport"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	412 #star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	413 #star_line = "Serotype Bardo shares the same antigenic profile with Newport, but Bardo is exceedingly rare."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	414 claim = "The serotype(s) is/are the only serotype(s) with the indicated antigenic profile currently recognized in the Kauffmann White Scheme. New serotypes can emerge and the possibility exists that this antigenic profile may emerge in a different subspecies. Identification of strains to the subspecies level should accompany serotype determination; the same antigenic profile in different subspecies is considered different serotypes.\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	415 if "N/A" in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	416 claim = ""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	417 #special test for Typhimurium
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	418 if "Typhimurium" in predict_sero or predict_form == "4:i:-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	419 normal = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	420 mutation = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	421 for x in special_gene_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	422 if "oafA-O-4_full" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	423 normal = float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	424 elif "oafA-O-4_5-" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	425 mutation = float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	426 if normal > mutation:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	427 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	428 elif normal < mutation:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	429 #predict_sero = predict_sero.strip() + "(O5-)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	430 predict_sero = predict_sero.strip() #diable special sero for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	431 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	432 #star_line = "Detected the deletion of O5-."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	433 star_line = "Detected a deletion that causes O5- variant of Typhimurium. "
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	434 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	435 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	436 #special test for Paratyphi B
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	437 if "Paratyphi B" in predict_sero or predict_form == "4:b:-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	438 normal = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	439 mutation = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	440 for x in special_gene_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	441 if "gntR-family-regulatory-protein_dt-positive" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	442 normal = float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	443 elif "gntR-family-regulatory-protein_dt-negative" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	444 mutation = float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	445 #print(normal,mutation)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	446 if normal > mutation:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	447 #predict_sero = predict_sero.strip() + "(dt+)" #diable special sero for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	448 predict_sero = predict_sero.strip()+' var. L(+) tartrate+' if "Paratyphi B" in predict_sero else predict_sero.strip()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	449 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	450 #star_line = "Didn't detect the SNP for dt- which means this isolate is a Paratyphi B variant L(+) tartrate(+)."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	451 star_line = "The SNP that causes d-Tartrate nonfermentating phenotype of Paratyphi B was not detected. "
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	452 elif normal < mutation:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	453 #predict_sero = predict_sero.strip() + "(dt-)" #diable special sero for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	454 predict_sero = predict_sero.strip()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	455 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	456 #star_line = "Detected the SNP for dt- which means this isolate is a systemic pathovar of Paratyphi B."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	457 star_line = "Detected the SNP for d-Tartrate nonfermenting phenotype of Paratyphi B. "
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	458 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	459 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	460 #star_line = " Failed to detect the SNP for dt-, can't decide it's a Paratyphi B variant L(+) tartrate(+) or not."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	461 star_line = " " ## ed_SL_05152019: do not report this situation.
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	462 #special test for O13,22 and O13,23
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	463 if Otype=="13":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	464 #ex_dir = os.path.dirname(os.path.realpath(__file__))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	465 ex_dir = os.path.abspath(os.path.join(os.path.dirname(os.path.dirname(__file__)),'seqsero2_db')) # ed_SL_09152019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	466 f = open(ex_dir + '/special.pickle', 'rb')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	467 special = pickle.load(f)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	468 O22_O23=special['O22_O23']
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	469 if predict_sero.split(" or ")[0] in O22_O23[-1] and predict_sero.split(" or ")[0] not in rename_dict_all:#if in rename_dict_all, then it means already merged, no need to analyze
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	470 O22_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	471 O23_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	472 for x in special_gene_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	473 if "O:22" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	474 O22_score = O22_score+float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	475 elif "O:23" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	476 O23_score = O23_score+float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	477 #print(O22_score,O23_score)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	478 for z in O22_O23[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	479 if predict_sero.split(" or ")[0] in z:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	480 if O22_score > O23_score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	481 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	482 #star_line = "Detected O22 specific genes to further differenciate '"+predict_sero+"'." #diabled for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	483 predict_sero = z[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	484 elif O22_score < O23_score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	485 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	486 #star_line = "Detected O23 specific genes to further differenciate '"+predict_sero+"'." #diabled for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	487 predict_sero = z[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	488 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	489 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	490 #star_line = "Fail to detect O22 and O23 differences." #diabled for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	491 if " or " in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	492 star_line = star_line + "The predicted serotypes share the same general formula: " + Otype + ":" + fliC + ":" + fljB + "."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	493 #special test for O6,8
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	494 #merge_O68_list=["Blockley","Bovismorbificans","Hadar","Litchfield","Manhattan","Muenchen"] #remove 11/11/2018, because already in merge list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	495 #for x in merge_O68_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	496 # if x in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	497 # predict_sero=x
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	498 # star=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	499 # star_line=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	500 #special test for Montevideo; most of them are monophasic
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	501 #if "Montevideo" in predict_sero and "1,2,7" in predict_form: #remove 11/11/2018, because already in merge list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	502 #star="*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	503 #star_line="Montevideo is almost always monophasic, having an antigen called for the fljB position may be a result of Salmonella-Salmonella contamination."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	504 return predict_form, predict_sero, star, star_line, claim
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	505 ### End of SeqSero Kmer part
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	506
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	507 ### Begin of SeqSero2 allele prediction and output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	508 def xml_parse_score_comparision_seqsero(xmlfile):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	509 #used to do seqsero xml analysis
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	510 from Bio.Blast import NCBIXML
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	511 handle=open(xmlfile)
12 08832c0d3cbd planemo upload commit 70dc513aa7d7ac6785847dfd86323687613b6b68-dirty cstrittmatter parents: 10 diff changeset	512 print(handle)
0 fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	513 handle=NCBIXML.parse(handle)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	514 handle=list(handle)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	515 List=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	516 List_score=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	517 List_ids=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	518 List_query_region=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	519 for i in range(len(handle)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	520 if len(handle[i].alignments)>0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	521 for j in range(len(handle[i].alignments)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	522 score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	523 ids=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	524 cover_region=set() #fixed problem that repeated calculation leading percentage > 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	525 List.append(handle[i].query.strip()+"___"+handle[i].alignments[j].hit_def)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	526 for z in range(len(handle[i].alignments[j].hsps)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	527 hsp=handle[i].alignments[j].hsps[z]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	528 temp=set(range(hsp.query_start,hsp.query_end))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	529 if len(cover_region)==0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	530 cover_region=cover_region\|temp
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	531 fraction=1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	532 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	533 fraction=1-len(cover_region&temp)/float(len(temp))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	534 cover_region=cover_region\|temp
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	535 if "last" in handle[i].query or "first" in handle[i].query:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	536 score+=hsp.bits*fraction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	537 ids+=float(hsp.identities)/handle[i].query_length*fraction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	538 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	539 score+=hsp.bits*fraction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	540 ids+=float(hsp.identities)/handle[i].query_length*fraction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	541 List_score.append(score)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	542 List_ids.append(ids)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	543 List_query_region.append(cover_region)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	544 temp=zip(List,List_score,List_ids,List_query_region)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	545 Final_list=sorted(temp, key=lambda d:d[1], reverse = True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	546 return Final_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	547
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	548
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	549 def Uniq(L,sort_on_fre="none"): #return the uniq list and the count number
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	550 Old=L
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	551 L.sort()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	552 L = [L[i] for i in range(len(L)) if L[i] not in L[:i]]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	553 count=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	554 for j in range(len(L)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	555 y=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	556 for x in Old:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	557 if L[j]==x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	558 y+=1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	559 count.append(y)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	560 if sort_on_fre!="none":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	561 d=zip(*sorted(zip(count, L)))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	562 L=d[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	563 count=d[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	564 return (L,count)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	565
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	566 def judge_fliC_or_fljB_from_head_tail_for_one_contig(nodes_vs_score_list):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	567 #used to predict it's fliC or fljB for one contig, based on tail and head score, but output the score difference,if it is very small, then not reliable, use blast score for whole contig to test
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	568 #this is mainly used for
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	569 a=nodes_vs_score_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	570 fliC_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	571 fljB_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	572 for z in a:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	573 if "fliC" in z[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	574 fliC_score+=z[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	575 elif "fljB" in z[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	576 fljB_score+=z[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	577 if fliC_score>=fljB_score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	578 role="fliC"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	579 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	580 role="fljB"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	581 return (role,abs(fliC_score-fljB_score))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	582
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	583 def judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(node_name,Final_list,Final_list_passed):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	584 #used to predict contig is fliC or fljB, if the differnce score value on above head_and_tail is less than 10 (quite small)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	585 #also used when no head or tail got blasted score for the contig
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	586 role=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	587 for z in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	588 if node_name in z[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	589 role=z[0].split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	590 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	591 return role
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	592
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	593 def fliC_or_fljB_judge_from_head_tail_sequence(nodes_list,tail_head_list,Final_list,Final_list_passed):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	594 #nodes_list is the c created by c,d=Uniq(nodes) in below function
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	595 first_target=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	596 role_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	597 for x in nodes_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	598 a=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	599 role=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	600 for y in tail_head_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	601 if x in y[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	602 a.append(y)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	603 if len(a)==4:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	604 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	605 if diff<20:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	606 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list,Final_list_passed)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	607 elif len(a)==3:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	608 ###however, if the one with highest score is the fewer one, compare their accumulation score
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	609 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	610 if diff<20:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	611 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list,Final_list_passed)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	612 ###end of above score comparison
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	613 elif len(a)==2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	614 #must on same node, if not, then decide with unit blast score, blast-score/length_of_special_sequence(30 or 37)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	615 temp=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	616 for z in a:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	617 temp.append(z[0].split("_")[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	618 m,n=Uniq(temp)#should only have one choice, but weird situation might occur too
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	619 if len(m)==1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	620 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	621 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	622 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	623 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	624 if diff<20:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	625 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list,Final_list_passed)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	626 ###need to desgin a algorithm to guess most possible situation for nodes_list, See the situations of test evaluation
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	627 elif len(a)==1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	628 #that one
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	629 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	630 if diff<20:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	631 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list,Final_list_passed)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	632 #need to evaluate, in future, may set up a cut-off, if not met, then just find Final_list_passed best match,like when "a==0"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	633 else:#a==0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	634 #use Final_list_passed best match
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	635 for z in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	636 if x in z[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	637 role=z[0].split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	638 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	639 #print x,role,len(a)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	640 role_list.append((role,x))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	641 if len(role_list)==2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	642 if role_list[0][0]==role_list[1][0]:#this is the most cocmmon error, two antigen were assigned to same phase
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	643 #just use score to do a final test
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	644 role_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	645 for x in nodes_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	646 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list,Final_list_passed)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	647 role_list.append((role,x))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	648 return role_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	649
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	650 def decide_contig_roles_for_H_antigen(Final_list,Final_list_passed):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	651 #used to decide which contig is FliC and which one is fljB
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	652 contigs=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	653 nodes=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	654 for x in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	655 if x[0].startswith("fl") and "last" not in x[0] and "first" not in x[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	656 nodes.append(x[0].split("___")[1].strip())
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	657 c,d=Uniq(nodes)#c is node_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	658 #print c
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	659 tail_head_list=[x for x in Final_list if ("last" in x[0] or "first" in x[0])]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	660 roles=fliC_or_fljB_judge_from_head_tail_sequence(c,tail_head_list,Final_list,Final_list_passed)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	661 return roles
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	662
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	663 def decide_O_type_and_get_special_genes(Final_list,Final_list_passed):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	664 #decide O based on Final_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	665 O_choice="?"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	666 O_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	667 special_genes={}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	668 nodes=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	669 for x in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	670 if x[0].startswith("O-"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	671 nodes.append(x[0].split("___")[1].strip())
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	672 elif not x[0].startswith("fl"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	673 special_genes[x[0]]=x[2]#08172018, x[2] changed from x[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	674 #print "special_genes:",special_genes
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	675 c,d=Uniq(nodes)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	676 #print "potential O antigen contig",c
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	677 final_O=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	678 O_nodes_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	679 for x in c:#c is the list for contigs
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	680 temp=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	681 for y in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	682 if x in y[0] and y[0].startswith("O-"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	683 final_O.append(y)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	684 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	685 ### O contig has the problem of two genes on same contig, so do additional test
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	686 potenial_new_gene=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	687 for x in final_O:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	688 pointer=0 #for genes merged or not
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	689 #not consider O-1,3,19_not_in_3,10, too short compared with others
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	690 if "O-1,3,19_not_in_3,10" not in x[0] and int(x[0].split("__")[1].split("___")[0])x[2]+850 <= int(x[0].split("length_")[1].split("_")[0]):#gene length << contig length; for now give 3002 (for secureity can use 400*2) as flank region
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	691 pointer=x[0].split("___")[1].strip()#store the contig name
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	692 print(pointer)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	693 if pointer!=0:#it has potential merge event
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	694 for y in Final_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	695 if pointer in y[0] and y not in final_O and (y[1]>=int(y[0].split("__")[1].split("___")[0])1.5 or (y[1]>=int(y[0].split("__")[1].split("___")[0])y[2] and y[1]>=400)):#that's a realtively strict filter now; if passed, it has merge event and add one more to final_O
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	696 potenial_new_gene=y
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	697 #print(potenial_new_gene)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	698 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	699 if potenial_new_gene!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	700 print("two differnt genes in same contig, fix it for O antigen")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	701 print(potenial_new_gene[:3])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	702 pointer=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	703 for y in final_O:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	704 if y[0].split("___")[-1]==potenial_new_gene[0].split("___")[-1]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	705 pointer=1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	706 if pointer!=0: #changed to consider two genes in same contig
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	707 final_O.append(potenial_new_gene)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	708 ### end of the two genes on same contig test
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	709 final_O=sorted(final_O,key=lambda x: x[2], reverse=True)#sorted
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	710 if len(final_O)==0 or (len(final_O)==1 and "O-1,3,19_not_in_3,10" in final_O[0][0]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	711 #print "$$$No Otype, due to no hit"#may need to be changed
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	712 O_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	713 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	714 highest_O_coverage=max([float(x[0].split("_cov_")[-1].split("_")[0]) for x in final_O if "O-1,3,19_not_in_3,10" not in x[0]])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	715 O_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	716 O_list_less_contamination=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	717 for x in final_O:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	718 if not "O-1,3,19_not_in_3,10__130" in x[0]:#O-1,3,19_not_in_3,10 is too small, which may affect further analysis; to avoid contamination affect, use 0.15 of highest coverage as cut-off
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	719 O_list.append(x[0].split("__")[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	720 O_nodes_list.append(x[0].split("___")[1])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	721 if float(x[0].split("_cov_")[-1].split("_")[0])>highest_O_coverage*0.15:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	722 O_list_less_contamination.append(x[0].split("__")[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	723 ### special test for O9,46 and O3,10 family
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	724 if ("O-9,46_wbaV" in O_list or "O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254" in O_list) and O_list_less_contamination[0].startswith("O-9,"):#not sure should use and float(O9_wbaV)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	725 if "O-9,46_wzy" in O_list or "O-9,46_wzy_partial" in O_list:#and float(O946_wzy)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	726 O_choice="O-9,46"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	727 #print "$$$Most possilble Otype: O-9,46"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	728 elif "O-9,46,27_partial_wzy" in O_list:#and float(O94627)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	729 O_choice="O-9,46,27"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	730 #print "$$$Most possilble Otype: O-9,46,27"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	731 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	732 O_choice="O-9"#next, detect O9 vs O2?
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	733 O2=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	734 O9=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	735 for z in special_genes:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	736 if "tyr-O-9" in z:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	737 O9=special_genes[z]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	738 elif "tyr-O-2" in z:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	739 O2=special_genes[z]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	740 if O2>O9:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	741 O_choice="O-2"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	742 elif O2<O9:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	743 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	744 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	745 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	746 #print "$$$No suitable one, because can't distinct it's O-9 or O-2, but O-9 has a more possibility."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	747 elif ("O-3,10_wzx" in O_list) and ("O-9,46_wzy" in O_list) and (O_list[0].startswith("O-3,10") or O_list_less_contamination[0].startswith("O-9,46_wzy")):#and float(O310_wzx)/float(num_1) > 0.1 and float(O946_wzy)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	748 if "O-3,10_not_in_1,3,19" in O_list:#and float(O310_no_1319)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	749 O_choice="O-3,10"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	750 #print "$$$Most possilble Otype: O-3,10 (contain O-3,10_not_in_1,3,19)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	751 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	752 O_choice="O-1,3,19"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	753 #print "$$$Most possilble Otype: O-1,3,19 (not contain O-3,10_not_in_1,3,19)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	754 ### end of special test for O9,46 and O3,10 family
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	755 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	756 try:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	757 max_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	758 for x in final_O:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	759 if x[2]>=max_score and float(x[0].split("_cov_")[-1].split("_")[0])>highest_O_coverage0.15:#use x[2],08172018, the "coverage identity = cover_length identity"; also meet coverage threshold
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	760 max_score=x[2]#change from x[-1] to x[2],08172018
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	761 O_choice=x[0].split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	762 if O_choice=="O-1,3,19":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	763 O_choice=final_O[1][0].split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	764 #print "$$$Most possilble Otype: ",O_choice
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	765 except:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	766 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	767 #print "$$$No suitable Otype, or failure of mapping (please check the quality of raw reads)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	768 if O_choice=="O-9,46,27" and len(O_list)==2 and "O-4_wzx" in O_list: #special for very low chance sitatuion between O4 and O9,27,46, this is for serotypes like Bredeney and Schwarzengrund (normallly O-4 will have higher score, but sometimes sequencing quality may affect the prediction)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	769 O_choice="O-4"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	770 #print "O:",O_choice,O_nodes_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	771 Otypes=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	772 for x in O_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	773 if x!="O-1,3,19_not_in_3,10":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	774 if "O-9,46_" not in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	775 Otypes.append(x.split("_")[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	776 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	777 Otypes.append(x.split("-from")[0])#O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	778 #Otypes=[x.split("_")[0] for x in O_list if x!="O-1,3,19_not_in_3,10"]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	779 Otypes_uniq,Otypes_fre=Uniq(Otypes)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	780 contamination_O=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	781 if O_choice=="O-9,46,27" or O_choice=="O-3,10" or O_choice=="O-1,3,19":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	782 if len(Otypes_uniq)>2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	783 contamination_O="potential contamination from O antigen signals"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	784 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	785 if len(Otypes_uniq)>1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	786 if O_choice=="O-4" and len(Otypes_uniq)==2 and "O-9,46,27" in Otypes_uniq: #for special 4,12,27 case such as Bredeney and Schwarzengrund
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	787 contamination_O=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	788 elif O_choice=="O-9,46" and len(Otypes_uniq)==2 and "O-9,46_wbaV" in Otypes_uniq and "O-9,46_wzy" in Otypes_uniq: #for special 4,12,27 case such as Bredeney and Schwarzengrund
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	789 contamination_O=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	790 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	791 contamination_O="potential contamination from O antigen signals"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	792 return O_choice,O_nodes_list,special_genes,final_O,contamination_O,Otypes_uniq
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	793 ### End of SeqSero2 allele prediction and output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	794
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	795 def get_input_files(make_dir,input_file,data_type,dirpath):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	796 #tell input files from datatype
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	797 #"<int>: '1'(pair-end reads, interleaved),'2'(pair-end reads, seperated),'3'(single-end reads), '4'(assembly),'5'(nanopore fasta),'6'(nanopore fastq)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	798 for_fq=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	799 rev_fq=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	800 os.chdir(make_dir)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	801 if data_type=="1":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	802 input_file=input_file[0].split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	803 if input_file.endswith(".sra"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	804 subprocess.check_call("fastq-dump --split-files "+input_file,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	805 for_fq=input_file.replace(".sra","_1.fastq")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	806 rev_fq=input_file.replace(".sra","_2.fastq")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	807 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	808 core_id=input_file.split(".fastq")[0].split(".fq")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	809 for_fq=core_id+"_1.fastq"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	810 rev_fq=core_id+"_2.fastq"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	811 if input_file.endswith(".gz"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	812 subprocess.check_call("gzip -dc "+input_file+" \| "+dirpath+"/deinterleave_fastq.sh "+for_fq+" "+rev_fq,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	813 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	814 subprocess.check_call("cat "+input_file+" \| "+dirpath+"/deinterleave_fastq.sh "+for_fq+" "+rev_fq,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	815 elif data_type=="2":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	816 for_fq=input_file[0].split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	817 rev_fq=input_file[1].split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	818 elif data_type=="3":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	819 input_file=input_file[0].split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	820 if input_file.endswith(".sra"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	821 subprocess.check_call("fastq-dump --split-files "+input_file,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	822 for_fq=input_file.replace(".sra","_1.fastq")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	823 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	824 for_fq=input_file
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	825 elif data_type in ["4","5","6"]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	826 for_fq=input_file[0].split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	827 os.chdir("..")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	828 return for_fq,rev_fq
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	829
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	830 def predict_O_and_H_types(Final_list,Final_list_passed,new_fasta):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	831 #get O and H types from Final_list from blast parsing; allele mode
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	832 from Bio import SeqIO
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	833 fliC_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	834 fljB_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	835 fliC_contig="NA"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	836 fljB_contig="NA"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	837 fliC_region=set([0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	838 fljB_region=set([0,])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	839 fliC_length=0 #can be changed to coverage in future; in 03292019, changed to ailgned length
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	840 fljB_length=0 #can be changed to coverage in future; in 03292019, changed to ailgned length
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	841 O_choice="-"#no need to decide O contig for now, should be only one
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	842 O_choice,O_nodes,special_gene_list,O_nodes_roles,contamination_O,Otypes_uniq=decide_O_type_and_get_special_genes(Final_list,Final_list_passed)#decide the O antigen type and also return special-gene-list for further identification
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	843 O_choice=O_choice.split("-")[-1].strip()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	844 if (O_choice=="1,3,19" and len(O_nodes_roles)==1 and "1,3,19" in O_nodes_roles[0][0]) or O_choice=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	845 O_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	846 H_contig_roles=decide_contig_roles_for_H_antigen(Final_list,Final_list_passed)#decide the H antigen contig is fliC or fljB
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	847 #add alignment locations, used for further selection, 03312019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	848 for i in range(len(H_contig_roles)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	849 x=H_contig_roles[i]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	850 for y in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	851 if x[1] in y[0] and y[0].startswith(x[0]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	852 H_contig_roles[i]+=H_contig_roles[i]+(y[-1],)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	853 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	854 log_file=open("SeqSero_log.txt","a")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	855 extract_file=open("Extracted_antigen_alleles.fasta","a")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	856 handle_fasta=list(SeqIO.parse(new_fasta,"fasta"))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	857
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	858 #print("O_contigs:")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	859 log_file.write("O_contigs:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	860 extract_file.write("#Sequences with antigen signals (if the micro-assembled contig only covers the flanking region, it will not be used for contamination analysis)\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	861 extract_file.write("#O_contigs:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	862 for x in O_nodes_roles:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	863 if "O-1,3,19_not_in_3,10" not in x[0]:#O-1,3,19_not_in_3,10 is just a small size marker
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	864 #print(x[0].split("___")[-1],x[0].split("__")[0],"blast score:",x[1],"identity%:",str(round(x[2]*100,2))+"%",str(min(x[-1]))+" to "+str(max(x[-1])))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	865 log_file.write(x[0].split("___")[-1]+" "+x[0].split("__")[0]+"; "+"blast score: "+str(x[1])+" identity%: "+str(round(x[2]*100,2))+"%; alignment from "+str(min(x[-1]))+" to "+str(max(x[-1]))+" of antigen\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	866 title=">"+x[0].split("___")[-1]+" "+x[0].split("__")[0]+"; "+"blast score: "+str(x[1])+" identity%: "+str(round(x[2]*100,2))+"%; alignment from "+str(min(x[-1]))+" to "+str(max(x[-1]))+" of antigen\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	867 seqs=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	868 for z in handle_fasta:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	869 if x[0].split("___")[-1]==z.description:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	870 seqs=str(z.seq)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	871 extract_file.write(title+seqs+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	872 if len(H_contig_roles)!=0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	873 highest_H_coverage=max([float(x[1].split("_cov_")[-1].split("_")[0]) for x in H_contig_roles]) #less than highest*0.1 would be regarded as contamination and noises, they will still be considered in contamination detection and logs, but not used as final serotype output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	874 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	875 highest_H_coverage=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	876 for x in H_contig_roles:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	877 #if multiple choices, temporately select the one with longest length for now, will revise in further change
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	878 if "fliC" == x[0] and len(x[-1])>=fliC_length and x[1] not in O_nodes and float(x[1].split("_cov_")[-1].split("_")[0])>highest_H_coverage*0.13:#remember to avoid the effect of O-type contig, so should not in O_node list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	879 fliC_contig=x[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	880 fliC_length=len(x[-1])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	881 elif "fljB" == x[0] and len(x[-1])>=fljB_length and x[1] not in O_nodes and float(x[1].split("_cov_")[-1].split("_")[0])>highest_H_coverage*0.13:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	882 fljB_contig=x[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	883 fljB_length=len(x[-1])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	884 for x in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	885 if fliC_choice=="-" and "fliC_" in x[0] and fliC_contig in x[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	886 fliC_choice=x[0].split("_")[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	887 elif fljB_choice=="-" and "fljB_" in x[0] and fljB_contig in x[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	888 fljB_choice=x[0].split("_")[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	889 elif fliC_choice!="-" and fljB_choice!="-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	890 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	891 #now remove contigs not in middle core part
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	892 first_allele="NA"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	893 first_allele_percentage=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	894 for x in Final_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	895 if x[0].startswith("fliC") or x[0].startswith("fljB"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	896 first_allele=x[0].split("__")[0] #used to filter those un-middle contigs
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	897 first_allele_percentage=x[2]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	898 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	899 additional_contigs=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	900 for x in Final_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	901 if first_allele in x[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	902 if (fliC_contig == x[0].split("___")[-1]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	903 fliC_region=x[3]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	904 elif fljB_contig!="NA" and (fljB_contig == x[0].split("___")[-1]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	905 fljB_region=x[3]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	906 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	907 if x[1]*1.1>int(x[0].split("___")[1].split("_")[3]):#loose threshold by multiplying 1.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	908 additional_contigs.append(x)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	909 #else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	910 #print x[:3]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	911 #we can just use the fljB region (or fliC depends on size), no matter set() or contain a large locations (without middle part); however, if none of them is fully assembled, use 500 and 1200 as conservative cut-off
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	912 if first_allele_percentage>0.9:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	913 if len(fliC_region)>len(fljB_region) and (max(fljB_region)-min(fljB_region))>1000:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	914 target_region=fljB_region\|(fliC_region-set(range(min(fljB_region),max(fljB_region)))) #fljB_region\|(fliC_region-set(range(min(fljB_region),max(fljB_region))))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	915 elif len(fliC_region)<len(fljB_region) and (max(fliC_region)-min(fliC_region))>1000:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	916 target_region=fliC_region\|(fljB_region-set(range(min(fliC_region),max(fliC_region)))) #fljB_region\|(fliC_region-set(range(min(fljB_region),max(fljB_region))))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	917 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	918 target_region=set()#doesn't do anything
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	919 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	920 target_region=set()#doesn't do anything
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	921 #print(target_region)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	922 #print(additional_contigs)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	923 target_region2=set(list(range(0,525))+list(range(1200,1700)))#I found to use 500 to 1200 as special region would be best
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	924 target_region=target_region2\|target_region
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	925 for x in additional_contigs:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	926 removal=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	927 contig_length=int(x[0].split("___")[1].split("length_")[-1].split("_")[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	928 if fljB_contig not in x[0] and fliC_contig not in x[0] and len(target_region&x[3])/float(len(x[3]))>0.65 and contig_length0.5<len(x[3])<contig_length1.5: #consider length and alignment length for now, but very loose,0.5 and 1.5 as cut-off
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	929 removal=1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	930 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	931 if first_allele_percentage > 0.9 and float(x[0].split("__")[1].split("___")[0])*x[2]/len(x[-1])>0.96:#if high similiarity with middle part of first allele (first allele >0.9, already cover middle part)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	932 removal=1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	933 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	934 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	935 if removal==1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	936 for y in H_contig_roles:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	937 if y[1] in x[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	938 H_contig_roles.remove(y)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	939 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	940 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	941 #print(x[:3],contig_length,len(target_region&x[3])/float(len(x[3])),contig_length0.5,len(x[3]),contig_length1.5)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	942 #end of removing none-middle contigs
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	943 #print("H_contigs:")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	944 log_file.write("H_contigs:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	945 extract_file.write("#H_contigs:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	946 H_contig_stat=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	947 H1_cont_stat={}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	948 H2_cont_stat={}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	949 for i in range(len(H_contig_roles)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	950 x=H_contig_roles[i]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	951 a=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	952 for y in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	953 if x[1] in y[0] and y[0].startswith(x[0]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	954 if "first" in y[0] or "last" in y[0]: #this is the final filter to decide it's fliC or fljB, if can't pass, then can't decide
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	955 for y in Final_list_passed: #it's impossible to has the "first" and "last" allele as prediction, so re-do it
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	956 if x[1] in y[0]:#it's very possible to be third phase allele, so no need to make it must be fliC or fljB
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	957 #print(x[1],"can't_decide_fliC_or_fljB",y[0].split("_")[1],"blast_score:",y[1],"identity%:",str(round(y[2]*100,2))+"%",str(min(y[-1]))+" to "+str(max(y[-1])))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	958 log_file.write(x[1]+" "+x[0]+" "+y[0].split("_")[1]+"; "+"blast score: "+str(y[1])+" identity%: "+str(round(y[2]*100,2))+"%; alignment from "+str(min(y[-1]))+" to "+str(max(y[-1]))+" of antigen\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	959 H_contig_roles[i]="can't decide fliC or fljB, may be third phase"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	960 title=">"+x[1]+" "+x[0]+" "+y[0].split("_")[1]+"; "+"blast score: "+str(y[1])+" identity%: "+str(round(y[2]*100,2))+"%; alignment from "+str(min(y[-1]))+" to "+str(max(y[-1]))+" of antiten\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	961 seqs=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	962 for z in handle_fasta:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	963 if x[1]==z.description:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	964 seqs=str(z.seq)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	965 extract_file.write(title+seqs+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	966 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	967 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	968 #print(x[1],x[0],y[0].split("_")[1],"blast_score:",y[1],"identity%:",str(round(y[2]*100,2))+"%",str(min(y[-1]))+" to "+str(max(y[-1])))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	969 log_file.write(x[1]+" "+x[0]+" "+y[0].split("_")[1]+"; "+"blast score: "+str(y[1])+" identity%: "+str(round(y[2]*100,2))+"%; alignment from "+str(min(y[-1]))+" to "+str(max(y[-1]))+" of antigen\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	970 title=">"+x[1]+" "+x[0]+" "+y[0].split("_")[1]+"; "+"blast score: "+str(y[1])+" identity%: "+str(round(y[2]*100,2))+"%; alignment from "+str(min(y[-1]))+" to "+str(max(y[-1]))+" of antigen\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	971 seqs=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	972 for z in handle_fasta:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	973 if x[1]==z.description:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	974 seqs=str(z.seq)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	975 extract_file.write(title+seqs+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	976 if x[0]=="fliC":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	977 if y[0].split("_")[1] not in H1_cont_stat:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	978 H1_cont_stat[y[0].split("_")[1]]=y[2]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	979 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	980 H1_cont_stat[y[0].split("_")[1]]+=y[2]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	981 if x[0]=="fljB":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	982 if y[0].split("_")[1] not in H2_cont_stat:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	983 H2_cont_stat[y[0].split("_")[1]]=y[2]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	984 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	985 H2_cont_stat[y[0].split("_")[1]]+=y[2]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	986 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	987 #detect contaminations
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	988 #print(H1_cont_stat)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	989 #print(H2_cont_stat)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	990 H1_cont_stat_list=[x for x in H1_cont_stat if H1_cont_stat[x]>0.2]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	991 H2_cont_stat_list=[x for x in H2_cont_stat if H2_cont_stat[x]>0.2]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	992 contamination_H=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	993 if len(H1_cont_stat_list)>1 or len(H2_cont_stat_list)>1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	994 contamination_H="potential contamination from H antigen signals"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	995 elif len(H2_cont_stat_list)==1 and fljB_contig=="NA":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	996 contamination_H="potential contamination from H antigen signals, uncommon weak fljB signals detected"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	997 #get additional antigens
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	998 """
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	999 if ("O-9,46_wbaV" in O_list or "O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254" in O_list) and O_list_less_contamination[0].startswith("O-9,"):#not sure should use and float(O9_wbaV)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1000 if "O-9,46_wzy" in O_list:#and float(O946_wzy)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1001 O_choice="O-9,46"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1002 #print "$$$Most possilble Otype: O-9,46"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1003 elif "O-9,46,27_partial_wzy" in O_list:#and float(O94627)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1004 O_choice="O-9,46,27"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1005 #print "$$$Most possilble Otype: O-9,46,27"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1006 elif ("O-3,10_wzx" in O_list) and ("O-9,46_wzy" in O_list) and (O_list[0].startswith("O-3,10") or O_list_less_contamination[0].startswith("O-9,46_wzy")):#and float(O310_wzx)/float(num_1) > 0.1 and float(O946_wzy)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1007 if "O-3,10_not_in_1,3,19" in O_list:#and float(O310_no_1319)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1008 O_choice="O-3,10"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1009 #print "$$$Most possilble Otype: O-3,10 (contain O-3,10_not_in_1,3,19)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1010 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1011 O_choice="O-1,3,19"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1012 #print "$$$Most possilble Otype: O-1,3,19 (not contain O-3,10_not_in_1,3,19)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1013 ### end of special test for O9,46 and O3,10 family
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1014
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1015 if O_choice=="O-9,46,27" or O_choice=="O-3,10" or O_choice=="O-1,3,19":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1016 if len(Otypes_uniq)>2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1017 contamination_O="potential contamination from O antigen signals"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1018 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1019 if len(Otypes_uniq)>1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1020 if O_choice=="O-4" and len(Otypes_uniq)==2 and "O-9,46,27" in Otypes_uniq: #for special 4,12,27 case such as Bredeney and Schwarzengrund
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1021 contamination_O=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1022 elif O_choice=="O-9,46" and len(Otypes_uniq)==2 and "O-9,46_wbaV" in Otypes_uniq and "O-9,46_wzy" in Otypes_uniq: #for special 4,12,27 case such as Bredeney and Schwarzengrund
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1023 contamination_O=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1024 """
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1025 additonal_antigents=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1026 #print(contamination_O)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1027 #print(contamination_H)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1028 log_file.write(contamination_O+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1029 log_file.write(contamination_H+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1030 log_file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1031 return O_choice,fliC_choice,fljB_choice,special_gene_list,contamination_O,contamination_H,Otypes_uniq,H1_cont_stat_list,H2_cont_stat_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1032
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1033 def get_input_K(input_file,lib_dict,data_type,k_size):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1034 #kmer mode; get input_Ks from dict and data_type
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1035 kmers = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1036 for h in lib_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1037 kmers += lib_dict[h]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1038 if data_type == '4':
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1039 input_Ks = target_multifasta_kmerizer(input_file, k_size, set(kmers))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1040 elif data_type == '1' or data_type == '2' or data_type == '3':#set it for now, will change later
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1041 input_Ks = target_read_kmerizer(input_file, k_size, set(kmers))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1042 elif data_type == '5':#minion_2d_fasta
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1043 input_Ks = minion_fasta_kmerizer(input_file, k_size, set(kmers))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1044 if data_type == '6':#minion_2d_fastq
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1045 input_Ks = minion_fastq_kmerizer(input_file, k_size, set(kmers))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1046 return input_Ks
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1047
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1048 def get_kmer_dict(lib_dict,input_Ks):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1049 #kmer mode; get predicted types
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1050 O_dict = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1051 H_dict = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1052 Special_dict = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1053 for h in lib_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1054 score = (len(lib_dict[h] & input_Ks) / len(lib_dict[h])) * 100
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1055 if score > 1: # Arbitrary cut-off for similarity score very low but seems necessary to detect O-3,10 in some cases
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1056 if h.startswith('O-') and score > 25:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1057 O_dict[h] = score
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1058 if h.startswith('fl') and score > 40:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1059 H_dict[h] = score
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1060 if (h[:2] != 'fl') and (h[:2] != 'O-'):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1061 Special_dict[h] = score
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1062 return O_dict,H_dict,Special_dict
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1063
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1064 def call_O_and_H_type(O_dict,H_dict,Special_dict,make_dir):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1065 log_file=open("SeqSero_log.txt","a")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1066 log_file.write("O_scores:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1067 #call O:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1068 highest_O = '-'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1069 if len(O_dict) == 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1070 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1071 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1072 for x in O_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1073 log_file.write(x+"\t"+str(O_dict[x])+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1074 if ('O-9,46_wbaV__1002' in O_dict and O_dict['O-9,46_wbaV__1002']>70) or ("O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254__1002" in O_dict and O_dict['O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254__1002']>70): # not sure should use and float(O9_wbaV)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1075 #if 'O-9,46_wzy__1191' in O_dict or "O-9,46_wzy_partial__216" in O_dict: # and float(O946_wzy)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1076 #modified to fix miscall of O-9,46
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1077 if ('O-9,46_wzy__1191' in O_dict and O_dict['O-9,46_wzy__1191']>40) or ("O-9,46_wzy_partial__216" in O_dict and O_dict["O-9,46_wzy_partial__216"]>40): # and float(O946_wzy)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1078 highest_O = "O-9,46"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1079 elif "O-9,46,27_partial_wzy__1019" in O_dict: # and float(O94627)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1080 highest_O = "O-9,46,27"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1081 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1082 highest_O = "O-9" # next, detect O9 vs O2?
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1083 O2 = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1084 O9 = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1085 for z in Special_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1086 if "tyr-O-9" in z:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1087 O9 = float(Special_dict[z])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1088 if "tyr-O-2" in z:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1089 O2 = float(Special_dict[z])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1090 if O2 > O9:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1091 highest_O = "O-2"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1092 elif ("O-3,10_wzx__1539" in O_dict) and (
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1093 "O-9,46_wzy__1191" in O_dict
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1094 ): # and float(O310_wzx)/float(num_1) > 0.1 and float(O946_wzy)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1095 if "O-3,10_not_in_1,3,19__1519" in O_dict: # and float(O310_no_1319)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1096 highest_O = "O-3,10"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1097 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1098 highest_O = "O-1,3,19"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1099 ### end of special test for O9,46 and O3,10 family
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1100 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1101 try:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1102 max_score = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1103 for x in O_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1104 if float(O_dict[x]) >= max_score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1105 max_score = float(O_dict[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1106 #highest_O = x.split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1107 # ed_SL_12182019: modified to fix the O-9,46 error example1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1108 if (x == 'O-9,46_wbaV__1002' or x == 'O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254__1002') and ('O-9,46_wzy__1191' not in O_dict and 'O-9,46_wzy_partial__216' not in O_dict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1109 highest_O = "O-9"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1110 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1111 highest_O = x.split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1112 if highest_O == "O-1,3,19":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1113 highest_O = '-'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1114 max_score = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1115 for x in O_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1116 if x == 'O-1,3,19_not_in_3,10__130':
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1117 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1118 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1119 if float(O_dict[x]) >= max_score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1120 max_score = float(O_dict[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1121 #highest_O = x.split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1122 # ed_SL_12182019: modified to fix the O-9,46 error example1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1123 if (x == 'O-9,46_wbaV__1002' or x == 'O-9,46_wbaV-from-II-9,12:z29:1,5-SRR1346254__1002') and ('O-9,46_wzy__1191' not in O_dict and 'O-9,46_wzy_partial__216' not in O_dict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1124 highest_O = "O-9"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1125 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1126 highest_O = x.split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1127 except:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1128 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1129 #call_fliC:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1130 if len(H_dict)!=0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1131 highest_H_score_both_BC=H_dict[max(H_dict.keys(), key=(lambda k: H_dict[k]))] #used to detect whether fljB existed or not
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1132 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1133 highest_H_score_both_BC=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1134 highest_fliC = '-'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1135 highest_fliC_raw = '-'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1136 highest_Score = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1137 log_file.write("\nH_scores:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1138 for s in H_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1139 log_file.write(s+"\t"+str(H_dict[s])+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1140 if s.startswith('fliC'):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1141 if float(H_dict[s]) > highest_Score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1142 highest_fliC = s.split('_')[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1143 highest_fliC_raw = s
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1144 highest_Score = float(H_dict[s])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1145 #call_fljB
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1146 highest_fljB = '-'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1147 highest_fljB_raw = '-'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1148 highest_Score = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1149 for s in H_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1150 if s.startswith('fljB'):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1151 if float(H_dict[s]) > highest_Score and float(H_dict[s]) > highest_H_score_both_BC * 0.65: #fljB is special, so use highest_H_score_both_BC to give a general estimate of coverage, currently 0.65 seems pretty good; the reason use a high (0.65) is some fliC and fljB shared with each other
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1152 #highest_fljB = s.split('_')[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1153 #highest_fljB_raw = s
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1154 #highest_Score = float(H_dict[s])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1155 if s.split('_')[1]!=highest_fliC:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1156 highest_fljB = s.split('_')[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1157 highest_fljB_raw = s
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1158 highest_Score = float(H_dict[s])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1159 log_file.write("\nSpecial_scores:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1160 for s in Special_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1161 log_file.write(s+"\t"+str(Special_dict[s])+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1162 log_file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1163 return highest_O,highest_fliC,highest_fljB
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1164
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1165 def get_temp_file_names(for_fq,rev_fq):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1166 #seqsero2 -a; get temp file names
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1167 sam=for_fq+".sam"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1168 bam=for_fq+".bam"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1169 sorted_bam=for_fq+"_sorted.bam"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1170 mapped_fq1=for_fq+"_mapped.fq"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1171 mapped_fq2=rev_fq+"_mapped.fq"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1172 combined_fq=for_fq+"_combined.fq"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1173 for_sai=for_fq+".sai"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1174 rev_sai=rev_fq+".sai"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1175 return sam,bam,sorted_bam,mapped_fq1,mapped_fq2,combined_fq,for_sai,rev_sai
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1176
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1177 def map_and_sort(threads,database,fnameA,fnameB,sam,bam,for_sai,rev_sai,sorted_bam,mapping_mode):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1178 #seqsero2 -a; do mapping and sort
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1179 print("building database...")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1180 subprocess.check_call("bwa index "+database+ " 2>> data_log.txt",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1181 print("mapping...")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1182 if mapping_mode=="mem":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1183 subprocess.check_call("bwa mem -k 17 -t "+threads+" "+database+" "+fnameA+" "+fnameB+" > "+sam+ " 2>> data_log.txt",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1184 elif mapping_mode=="sam":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1185 if fnameB!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1186 subprocess.check_call("bwa aln -t "+threads+" "+database+" "+fnameA+" > "+for_sai+ " 2>> data_log.txt",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1187 subprocess.check_call("bwa aln -t "+threads+" "+database+" "+fnameB+" > "+rev_sai+ " 2>> data_log.txt",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1188 subprocess.check_call("bwa sampe "+database+" "+for_sai+" "+ rev_sai+" "+fnameA+" "+fnameB+" > "+sam+ " 2>> data_log.txt",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1189 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1190 subprocess.check_call("bwa aln -t "+threads+" "+database+" "+fnameA+" > "+for_sai+ " 2>> data_log.txt",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1191 subprocess.check_call("bwa samse "+database+" "+for_sai+" "+for_fq+" > "+sam)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1192 subprocess.check_call("samtools view -@ "+threads+" -F 4 -Sh "+sam+" > "+bam,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1193 ### check the version of samtools then use differnt commands
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1194 samtools_version=subprocess.Popen(["samtools"],stdout=subprocess.PIPE,stderr=subprocess.PIPE)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1195 out, err = samtools_version.communicate()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1196 version = str(err).split("ersion:")[1].strip().split(" ")[0].strip()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1197 print("check samtools version:",version)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1198 ### end of samtools version check and its analysis
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1199 if LooseVersion(version)<=LooseVersion("1.2"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1200 subprocess.check_call("samtools sort -@ "+threads+" -n "+bam+" "+fnameA+"_sorted",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1201 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1202 subprocess.check_call("samtools sort -@ "+threads+" -n "+bam+" >"+sorted_bam,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1203
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1204 def extract_mapped_reads_and_do_assembly_and_blast(current_time,sorted_bam,combined_fq,mapped_fq1,mapped_fq2,threads,fnameA,fnameB,database,mapping_mode):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1205 #seqsero2 -a; extract, assembly and blast
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1206 subprocess.check_call("bamToFastq -i "+sorted_bam+" -fq "+combined_fq,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1207 #print("fnameA:",fnameA)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1208 #print("fnameB:",fnameB)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1209 if fnameB!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1210 subprocess.check_call("bamToFastq -i "+sorted_bam+" -fq "+mapped_fq1+" -fq2 "+mapped_fq2 + " 2>> data_log.txt",shell=True)#2> /dev/null if want no output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1211 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1212 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1213 outdir=current_time+"_temp"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1214 print("assembling...")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1215 if int(threads)>4:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1216 t="4"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1217 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1218 t=threads
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1219 if os.path.getsize(combined_fq)>100 and (fnameB=="" or os.path.getsize(mapped_fq1)>100):#if not, then it's "-:-:-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1220 if fnameB!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1221 subprocess.check_call("spades.py --careful --pe1-s "+combined_fq+" --pe1-1 "+mapped_fq1+" --pe1-2 "+mapped_fq2+" -t "+t+" -o "+outdir+ " >> data_log.txt 2>&1",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1222 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1223 subprocess.check_call("spades.py --careful --pe1-s "+combined_fq+" -t "+t+" -o "+outdir+ " >> data_log.txt 2>&1",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1224 new_fasta=fnameA+"_"+database+"_"+mapping_mode+".fasta"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1225 #new_fasta=fnameA+"_"+database.split('/')[-1]+"_"+mapping_mode+".fasta" # change path to databse for packaging
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1226 subprocess.check_call("mv "+outdir+"/contigs.fasta "+new_fasta+ " 2> /dev/null",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1227 #os.system("mv "+outdir+"/scaffolds.fasta "+new_fasta+ " 2> /dev/null") contigs.fasta
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1228 subprocess.check_call("rm -rf "+outdir+ " 2> /dev/null",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1229 print("blasting...","\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1230 xmlfile="blasted_output.xml"#fnameA+"-extracted_vs_"+database+"_"+mapping_mode+".xml"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1231 subprocess.check_call('makeblastdb -in '+new_fasta+' -out '+new_fasta+'_db '+'-dbtype nucl >> data_log.txt 2>&1',shell=True) #temp.txt is to forbid the blast result interrupt the output of our program###1/27/2015
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1232 subprocess.check_call("blastn -query "+database+" -db "+new_fasta+"_db -out "+xmlfile+" -outfmt 5 >> data_log.txt 2>&1",shell=True)###1/27/2015; 08272018, remove "-word_size 10"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1233 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1234 xmlfile="NA"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1235 return xmlfile,new_fasta
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1236
7 aa54a94b9aeb planemo upload commit c50df40caef2fb97c178d6890961e0e527992324-dirty cstrittmatter parents: 0 diff changeset	1237 def judge_subspecies(fnameA,dirpath):
0 fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1238 #seqsero2 -a; judge subspecies on just forward raw reads fastq
7 aa54a94b9aeb planemo upload commit c50df40caef2fb97c178d6890961e0e527992324-dirty cstrittmatter parents: 0 diff changeset	1239 samid_strcmd = "python " + dirpath + "/../SalmID.py -i "+fnameA
aa54a94b9aeb planemo upload commit c50df40caef2fb97c178d6890961e0e527992324-dirty cstrittmatter parents: 0 diff changeset	1240 print(samid_strcmd)
10 e6437d423693 planemo upload commit 70dc513aa7d7ac6785847dfd86323687613b6b68-dirty cstrittmatter parents: 8 diff changeset	1241 #seqsero2 -a; judge subspecies on just forward raw reads fastq
e6437d423693 planemo upload commit 70dc513aa7d7ac6785847dfd86323687613b6b68-dirty cstrittmatter parents: 8 diff changeset	1242 #salmID_output=subprocess.Popen("SalmID.py -i "+fnameA,shell=True,stdout=subprocess.PIPE,stderr=subprocess.PIPE)
7 aa54a94b9aeb planemo upload commit c50df40caef2fb97c178d6890961e0e527992324-dirty cstrittmatter parents: 0 diff changeset	1243 salmID_output=subprocess.Popen(samid_strcmd,shell=True,stdout=subprocess.PIPE,stderr=subprocess.PIPE)
0 fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1244 out, err = salmID_output.communicate()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1245 out=out.decode("utf-8")
10 e6437d423693 planemo upload commit 70dc513aa7d7ac6785847dfd86323687613b6b68-dirty cstrittmatter parents: 8 diff changeset	1246 print(out)
e6437d423693 planemo upload commit 70dc513aa7d7ac6785847dfd86323687613b6b68-dirty cstrittmatter parents: 8 diff changeset	1247 print(err)
0 fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1248 file=open("data_log.txt","a")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1249 file.write(out)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1250 file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1251 salm_species_scores=out.split("\n")[1].split("\t")[6:]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1252 salm_species_results=out.split("\n")[0].split("\t")[6:]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1253 max_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1254 max_score_index=1 #default is 1, means "I"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1255 for i in range(len(salm_species_scores)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1256 if max_score<float(salm_species_scores[i]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1257 max_score=float(salm_species_scores[i])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1258 max_score_index=i
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1259 prediction=salm_species_results[max_score_index].split(".")[1].strip().split(" ")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1260 #if float(out.split("\n")[1].split("\t")[4]) > float(out.split("\n")[1].split("\t")[5]): #bongori and enterica compare
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1261 if float(out.split("\n")[1].split("\t")[4]) > 10 and float(out.split("\n")[1].split("\t")[4]) > float(out.split("\n")[1].split("\t")[5]): ## ed_SL_0318: change SalmID_ssp_threshold
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1262 prediction="bongori" #if not, the prediction would always be enterica, since they are located in the later part
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1263 #if max_score<10: ## ed_SL_0318: change SalmID_ssp_threshold
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1264 if max_score<60:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1265 prediction="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1266 return prediction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1267
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1268 def judge_subspecies_Kmer(Special_dict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1269 #seqsero2 -k;
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1270 max_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1271 prediction="-" #default should be I
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1272 for x in Special_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1273 #if "mer" in x: ## ed_SL_0318: change ssp_threshold
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1274 if "mer" in x and float(Special_dict[x]) > 60:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1275 if max_score<float(Special_dict[x]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1276 max_score=float(Special_dict[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1277 prediction=x.split("_")[-1].strip()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1278 if x.split("_")[-1].strip()=="bongori" and float(Special_dict[x])>95:#if bongori already, then no need to test enterica
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1279 prediction="bongori"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1280 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1281 return prediction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1282
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1283 ## ed_SL_11232019: add notes for missing antigen
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1284 def check_antigens(ssp,O_antigen,H1_antigen,H2_antigen,NA_note):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1285 antigen_note = ''
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1286 if ssp != '-':
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1287 if O_antigen != '-' and H1_antigen == '-' and H2_antigen == '-': # O:-:-
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1288 antigen_note = 'H antigens were not detected. This is an atypical result that should be further investigated. Most Salmonella strains have at least fliC, encoding the Phase 1 H antigen, even if it is not expressed. '
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1289 NA_note = ''
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1290 elif O_antigen != '-' and H1_antigen == '-' and H2_antigen != '-': # O:-:H2
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1291 antigen_note = 'fliC was not detected. This is an atypical result that should be further investigated. Most Salmonella strains have fliC, encoding the Phase 1 H antigen, even if it is not expressed. '
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1292 NA_note = ''
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1293 elif O_antigen == '-' and H1_antigen != '-': # -:H1:X
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1294 antigen_note = 'O antigen was not detected. This result may be due to a rough strain that has deleted the rfb region. For raw reads input, the k-mer workflow is sometimes more sensitive than the microassembly workflow in detecting O antigen. Caution should be used with this approach because the k-mer result may be due to low levels of contamination. '
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1295 NA_note = ''
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1296 elif O_antigen == '-' and H1_antigen == '-' and H2_antigen == '-': # -:-:-
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1297 antigen_note = 'No serotype antigens were detected. This is an atypical result that should be further investigated. '
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1298 NA_note = ''
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1299 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1300 antigen_note = 'The input genome cannot be identified as Salmonella. Check the input for taxonomic ID, contamination, or sequencing quality. '
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1301 NA_note = ''
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1302 # if [O_antigen, H1_antigen, H2_antigen].count('-') >= 2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1303 # antigen_note = 'No subspecies marker was detected and less than 2 serotype antigens were detected; further, this genome was not identified as Salmonella. This is an atypical result that should be further investigated. '
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1304 # else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1305 # antigen_note = 'No subspecies marker was detected. This genome may not be Salmonella. This is an atypical result that should be further investigated. '
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1306 return (antigen_note,NA_note)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1307
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1308 def main():
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1309 #combine SeqSeroK and SeqSero2, also with SalmID
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1310 args = parse_args()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1311 input_file = args.i
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1312 data_type = args.t
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1313 analysis_mode = args.m
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1314 mapping_mode=args.b
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1315 threads=args.p
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1316 make_dir=args.d
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1317 clean_mode=args.c
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1318 sample_name=args.n
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1319 ingore_header=args.s
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1320 k_size=27 #will change for bug fixing
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1321 dirpath = os.path.abspath(os.path.dirname(os.path.realpath(__file__)))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1322 ex_dir = os.path.abspath(os.path.join(os.path.dirname(os.path.dirname(__file__)),'seqsero2_db')) # ed_SL_09152019: add ex_dir for packaging
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1323 seqsero2_db=ex_dir+"/H_and_O_and_specific_genes.fasta" # ed_SL_11092019: change path to database for packaging
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1324 database="H_and_O_and_specific_genes.fasta"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1325 note="Note: "
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1326 NA_note="This predicted serotype is not in the Kauffman-White scheme. " # ed_SL_09272019: add for new output format
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1327 if len(sys.argv)==1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1328 subprocess.check_call(dirpath+"/SeqSero2_package.py -h",shell=True)#change name of python file
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1329 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1330 request_id = time.strftime("%m_%d_%Y_%H_%M_%S", time.localtime())
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1331 request_id += str(random.randint(1, 10000000))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1332 if make_dir is None:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1333 make_dir="SeqSero_result_"+request_id
8 357e38526e2a planemo upload commit c50df40caef2fb97c178d6890961e0e527992324-dirty cstrittmatter parents: 7 diff changeset	1334 make_dir=os.path.abspath(make_dir)
0 fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1335 if os.path.isdir(make_dir):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1336 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1337 else:
8 357e38526e2a planemo upload commit c50df40caef2fb97c178d6890961e0e527992324-dirty cstrittmatter parents: 7 diff changeset	1338 subprocess.check_call("mkdir -p "+make_dir,shell=True)
0 fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1339 #subprocess.check_call("cp "+dirpath+"/"+database+" "+" ".join(input_file)+" "+make_dir,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1340 #subprocess.check_call("ln -sr "+dirpath+"/"+database+" "+" ".join(input_file)+" "+make_dir,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1341 subprocess.check_call("ln -f -s "+seqsero2_db+" "+" ".join(input_file)+" "+make_dir,shell=True) # ed_SL_11092019: change path to database for packaging
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1342 #subprocess.check_call("ln -f -s "+dirpath+"/"+database+" "+" ".join(input_file)+" "+make_dir,shell=True) ### use -f option to force the replacement of links, remove -r and use absolute path instead to avoid link issue (use 'type=os.path.abspath' in -i argument).
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1343 ############################begin the real analysis
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1344 if analysis_mode=="a":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1345 if data_type in ["1","2","3"]:#use allele mode
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1346 for_fq,rev_fq=get_input_files(make_dir,input_file,data_type,dirpath)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1347 os.chdir(make_dir)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1348 ###add a function to tell input files
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1349 fnameA=for_fq.split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1350 fnameB=rev_fq.split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1351 current_time=time.strftime("%Y_%m_%d_%H_%M_%S", time.localtime())
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1352 sam,bam,sorted_bam,mapped_fq1,mapped_fq2,combined_fq,for_sai,rev_sai=get_temp_file_names(fnameA,fnameB) #get temp files id
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1353 map_and_sort(threads,database,fnameA,fnameB,sam,bam,for_sai,rev_sai,sorted_bam,mapping_mode) #do mapping and sort
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1354 ### avoid error out when micro assembly fails. ed_SL_03172020
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1355 try:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1356 xmlfile,new_fasta=extract_mapped_reads_and_do_assembly_and_blast(current_time,sorted_bam,combined_fq,mapped_fq1,mapped_fq2,threads,fnameA,fnameB,database,mapping_mode) #extract the mapped reads and do micro assembly and blast
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1357 except (UnboundLocalError, subprocess.CalledProcessError):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1358 xmlfile="NA"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1359 H1_cont_stat_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1360 H2_cont_stat_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1361 ###
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1362 if xmlfile=="NA":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1363 O_choice,fliC_choice,fljB_choice,special_gene_list,contamination_O,contamination_H=("-","-","-",[],"","")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1364 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1365 Final_list=xml_parse_score_comparision_seqsero(xmlfile) #analyze xml and get parsed results
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1366 file=open("data_log.txt","a")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1367 for x in Final_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1368 file.write("\t".join(str(y) for y in x)+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1369 file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1370 Final_list_passed=[x for x in Final_list if float(x[0].split("_cov_")[1].split("_")[0])>=0.9 and (x[1]>=int(x[0].split("__")[1]) or x[1]>=int(x[0].split("___")[1].split("_")[3]) or x[1]>1000)]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1371 O_choice,fliC_choice,fljB_choice,special_gene_list,contamination_O,contamination_H,Otypes_uniq,H1_cont_stat_list,H2_cont_stat_list=predict_O_and_H_types(Final_list,Final_list_passed,new_fasta) #predict O, fliC and fljB
7 aa54a94b9aeb planemo upload commit c50df40caef2fb97c178d6890961e0e527992324-dirty cstrittmatter parents: 0 diff changeset	1372 subspecies=judge_subspecies(fnameA,dirpath) #predict subspecies
0 fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1373 ###output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1374 predict_form,predict_sero,star,star_line,claim=seqsero_from_formula_to_serotypes(O_choice,fliC_choice,fljB_choice,special_gene_list,subspecies)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1375 claim="" #04132019, disable claim for new report requirement
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1376 contamination_report=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1377 H_list=["fliC_"+x for x in H1_cont_stat_list if len(x)>0]+["fljB_"+x for x in H2_cont_stat_list if len(x)>0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1378 if contamination_O!="" and contamination_H=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1379 contamination_report="#Potential inter-serotype contamination detected from O antigen signals. All O-antigens detected:"+"\t".join(Otypes_uniq)+"."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1380 elif contamination_O=="" and contamination_H!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1381 contamination_report="#Potential inter-serotype contamination detected or potential thrid H phase from H antigen signals. All H-antigens detected:"+"\t".join(H_list)+"."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1382 elif contamination_O!="" and contamination_H!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1383 contamination_report="#Potential inter-serotype contamination detected from both O and H antigen signals.All O-antigens detected:"+"\t".join(Otypes_uniq)+". All H-antigens detected:"+"\t".join(H_list)+"."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1384 if contamination_report!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1385 #contamination_report="potential inter-serotype contamination detected (please refer below antigen signal report for details)." #above contamination_reports are for back-up and bug fixing #web-based mode need to be re-used, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1386 contamination_report="Co-existence of multiple serotypes detected, indicating potential inter-serotype contamination. See 'Extracted_antigen_alleles.fasta' for detected serotype determinant alleles. "
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1387 #claim="\n"+open("Extracted_antigen_alleles.fasta","r").read()#used to store H and O antigen sequeences #04132019, need to change if using web-version
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1388 #if contamination_report+star_line+claim=="": #0413, new output style
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1389 # note=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1390 #else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1391 # note="Note:"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1392
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1393 ### ed_SL_11232019: add notes for missing antigen
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1394 if O_choice=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1395 O_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1396 antigen_note,NA_note=check_antigens(subspecies,O_choice,fliC_choice,fljB_choice,NA_note)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1397 if sample_name:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1398 print ("Sample name:\t"+sample_name)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1399 ###
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1400
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1401 if clean_mode:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1402 subprocess.check_call("rm -rf ../"+make_dir,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1403 make_dir="none-output-directory due to '-c' flag"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1404 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1405 new_file=open("SeqSero_result.txt","w")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1406 ### ed_SL_01152020: add new output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1407 conta_note="yes" if "inter-serotype contamination" in contamination_report else "no"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1408 tsv_file=open("SeqSero_result.tsv","w")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1409 if ingore_header:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1410 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1411 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1412 tsv_file.write("Sample name\tOutput directory\tInput files\tO antigen prediction\tH1 antigen prediction(fliC)\tH2 antigen prediction(fljB)\tPredicted subspecies\tPredicted antigenic profile\tPredicted serotype\tPotential inter-serotype contamination\tNote\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1413 if sample_name:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1414 new_file.write("Sample name:\t"+sample_name+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1415 tsv_file.write(sample_name+'\t')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1416 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1417 tsv_file.write(input_file[0].split('/')[-1]+'\t')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1418 ###
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1419 if "N/A" not in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1420 new_file.write("Output directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1421 "Input files:\t"+"\t".join(input_file)+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1422 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1423 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1424 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1425 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1426 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1427 "Predicted serotype:\t"+predict_sero+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1428 note+contamination_report+star_line+claim+antigen_note+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1429 tsv_file.write(make_dir+"\t"+" ".join(input_file)+"\t"+O_choice+"\t"+fliC_choice+"\t"+fljB_choice+"\t"+subspecies+"\t"+predict_form+"\t"+predict_sero+"\t"+conta_note+"\t"+contamination_report+star_line+claim+antigen_note+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1430 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1431 #star_line=star_line.strip()+"\tNone such antigenic formula in KW.\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1432 star_line="" #04132019, for new output requirement, diable star_line if "NA" in output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1433 new_file.write("Output directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1434 "Input files:\t"+"\t".join(input_file)+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1435 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1436 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1437 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1438 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1439 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1440 "Predicted serotype:\t"+subspecies+' '+predict_form+"\n"+ # add serotype output for "N/A" prediction, add subspecies
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1441 note+NA_note+contamination_report+star_line+claim+antigen_note+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1442 tsv_file.write(make_dir+"\t"+" ".join(input_file)+"\t"+O_choice+"\t"+fliC_choice+"\t"+fljB_choice+"\t"+subspecies+"\t"+predict_form+"\t"+subspecies+' '+predict_form+"\t"+conta_note+"\t"+NA_note+contamination_report+star_line+claim+antigen_note+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1443 new_file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1444 tsv_file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1445 #subprocess.check_call("cat Seqsero_result.txt",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1446 #subprocess.call("rm H_and_O_and_specific_genes.fasta* .sra .bam .sam .fastq .gz .fq temp.txt .xml "+fnameA+"_db* 2> /dev/null",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1447 subprocess.call("rm H_and_O_and_specific_genes.fasta* .sra .bam .sam .fastq .gz .fq temp.txt "+fnameA+"_db 2> /dev/null",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1448 if "N/A" not in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1449 #print("Output_directory:"+make_dir+"\nInput files:\t"+for_fq+" "+rev_fq+"\n"+"O antigen prediction:\t"+O_choice+"\n"+"H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+"H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+"Predicted antigenic profile:\t"+predict_form+"\n"+"Predicted subspecies:\t"+subspecies+"\n"+"Predicted serotype(s):\t"+predict_sero+star+"\nNote:"+contamination_report+star+star_line+claim+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1450 print("Output directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1451 "Input files:\t"+"\t".join(input_file)+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1452 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1453 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1454 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1455 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1456 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1457 "Predicted serotype:\t"+predict_sero+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1458 note+contamination_report+star_line+claim+antigen_note+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1459 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1460 print("Output directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1461 "Input files:\t"+"\t".join(input_file)+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1462 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1463 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1464 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1465 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1466 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1467 "Predicted serotype:\t"+subspecies+' '+predict_form+"\n"+ # add serotype output for "N/A" prediction, subspecies
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1468 note+NA_note+contamination_report+star_line+claim+antigen_note+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1469 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1470 print("Allele modes only support raw reads datatype, i.e. '-t 1 or 2 or 3'; please use '-m k'")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1471 elif analysis_mode=="k":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1472 #ex_dir = os.path.dirname(os.path.realpath(__file__))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1473 ex_dir = os.path.abspath(os.path.join(os.path.dirname(os.path.dirname(__file__)),'seqsero2_db')) # ed_SL_09152019: change ex_dir for packaging
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1474 #output_mode = args.mode
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1475 for_fq,rev_fq=get_input_files(make_dir,input_file,data_type,dirpath)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1476 input_file = for_fq #-k will just use forward because not all reads were used
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1477 os.chdir(make_dir)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1478 f = open(ex_dir + '/antigens.pickle', 'rb')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1479 lib_dict = pickle.load(f)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1480 f.close
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1481 input_Ks=get_input_K(input_file,lib_dict,data_type,k_size)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1482 O_dict,H_dict,Special_dict=get_kmer_dict(lib_dict,input_Ks)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1483 highest_O,highest_fliC,highest_fljB=call_O_and_H_type(O_dict,H_dict,Special_dict,make_dir)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1484 subspecies=judge_subspecies_Kmer(Special_dict)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1485 if subspecies=="IIb" or subspecies=="IIa":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1486 subspecies="II"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1487 predict_form,predict_sero,star,star_line,claim = seqsero_from_formula_to_serotypes(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1488 highest_O.split('-')[1], highest_fliC, highest_fljB, Special_dict,subspecies)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1489 claim="" #no claim any more based on new output requirement
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1490 #if star_line+claim=="": #0413, new output style
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1491 # note=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1492 #else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1493 # note="Note:"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1494
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1495 ### ed_SL_11232019: add notes for missing antigen
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1496 if highest_O.split('-')[-1]=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1497 O_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1498 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1499 O_choice=highest_O.split('-')[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1500 antigen_note,NA_note=check_antigens(subspecies,O_choice,highest_fliC,highest_fljB,NA_note)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1501 if sample_name:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1502 print ("Sample name:\t"+sample_name)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1503 ###
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1504
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1505 if clean_mode:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1506 subprocess.check_call("rm -rf ../"+make_dir,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1507 make_dir="none-output-directory due to '-c' flag"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1508 # ### ed_SL_05282019, fix the assignment issue of variable 'O_choice' using "-m k -c"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1509 # if highest_O.split('-')[-1]=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1510 # O_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1511 # else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1512 # O_choice=highest_O.split('-')[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1513 # ###
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1514 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1515 # if highest_O.split('-')[-1]=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1516 # O_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1517 # else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1518 # O_choice=highest_O.split('-')[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1519 #print("Output_directory:"+make_dir+"\tInput_file:"+input_file+"\tPredicted subpecies:"+subspecies + '\tPredicted antigenic profile:' + predict_form + '\tPredicted serotype(s):' + predict_sero)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1520 new_file=open("SeqSero_result.txt","w")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1521 #new_file.write("Output_directory:"+make_dir+"\nInput files:\t"+input_file+"\n"+"O antigen prediction:\t"+O_choice+"\n"+"H1 antigen prediction(fliC):\t"+highest_fliC+"\n"+"H2 antigen prediction(fljB):\t"+highest_fljB+"\n"+"Predicted antigenic profile:\t"+predict_form+"\n"+"Predicted subspecies:\t"+subspecies+"\n"+"Predicted serotype(s):\t"+predict_sero+star+"\n"+star+star_line+claim+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1522 ### ed_SL_01152020: add new output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1523 tsv_file=open("SeqSero_result.tsv","w")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1524 if ingore_header:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1525 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1526 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1527 tsv_file.write("Sample name\tOutput directory\tInput files\tO antigen prediction\tH1 antigen prediction(fliC)\tH2 antigen prediction(fljB)\tPredicted subspecies\tPredicted antigenic profile\tPredicted serotype\tNote\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1528 if sample_name:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1529 new_file.write("Sample name:\t"+sample_name+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1530 tsv_file.write(sample_name+'\t')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1531 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1532 tsv_file.write(input_file.split('/')[-1]+'\t')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1533 ###
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1534 if "N/A" not in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1535 new_file.write("Output directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1536 "Input files:\t"+input_file+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1537 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1538 "H1 antigen prediction(fliC):\t"+highest_fliC+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1539 "H2 antigen prediction(fljB):\t"+highest_fljB+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1540 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1541 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1542 "Predicted serotype:\t"+predict_sero+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1543 note+star_line+claim+antigen_note+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1544 tsv_file.write(make_dir+"\t"+input_file+"\t"+O_choice+"\t"+highest_fliC+"\t"+highest_fljB+"\t"+subspecies+"\t"+predict_form+"\t"+predict_sero+"\t"+star_line+claim+antigen_note+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1545 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1546 #star_line=star_line.strip()+"\tNone such antigenic formula in KW.\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1547 star_line = "" #changed for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1548 new_file.write("Output directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1549 "Input files:\t"+input_file+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1550 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1551 "H1 antigen prediction(fliC):\t"+highest_fliC+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1552 "H2 antigen prediction(fljB):\t"+highest_fljB+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1553 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1554 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1555 "Predicted serotype:\t"+subspecies+' '+predict_form+"\n"+ # add serotype output for "N/A" prediction, subspecies
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1556 note+NA_note+star_line+claim+antigen_note+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1557 tsv_file.write(make_dir+"\t"+input_file+"\t"+O_choice+"\t"+highest_fliC+"\t"+highest_fljB+"\t"+subspecies+"\t"+predict_form+"\t"+subspecies+' '+predict_form+"\t"+NA_note+star_line+claim+antigen_note+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1558 new_file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1559 tsv_file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1560 subprocess.call("rm .fasta .fastq .gz .fq temp.txt .sra 2> /dev/null",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1561 if "N/A" not in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1562 print("Output directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1563 "Input files:\t"+input_file+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1564 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1565 "H1 antigen prediction(fliC):\t"+highest_fliC+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1566 "H2 antigen prediction(fljB):\t"+highest_fljB+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1567 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1568 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1569 "Predicted serotype:\t"+predict_sero+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1570 note+star_line+claim+antigen_note+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1571 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1572 print("Output directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1573 "Input files:\t"+input_file+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1574 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1575 "H1 antigen prediction(fliC):\t"+highest_fliC+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1576 "H2 antigen prediction(fljB):\t"+highest_fljB+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1577 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1578 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1579 "Predicted serotype:\t"+subspecies+' '+predict_form+"\n"+ # add serotype output for "N/A" prediction, subspecies
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1580 note+NA_note+star_line+claim+antigen_note+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1581
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1582 if __name__ == '__main__':
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1583 main()

Mercurial > repos > cstrittmatter > ss2v110

annotate bin/SeqSero2_package.py @ 14:047956dacae4 draft default tip