ss2v110: SeqSero2_package.py annotate

annotate SeqSero2_package.py @ 9:43f6b7f6ebb3 draft

planemo upload commit c50df40caef2fb97c178d6890961e0e527992324-dirty

author	cstrittmatter
date	Thu, 30 Apr 2020 21:47:42 -0400
parents	fc22ec8e924e
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 import math
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	14 from distutils.version import LooseVersion
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	15 from distutils.spawn import find_executable
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."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: 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: v1.0.2')#add "-m <data_type>" in future
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) ### ed_SL_05282019: 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("-d",help="<string>: 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	32 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	33 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	34 parser.add_argument('-v', '--version', action='version', version='%(prog)s ' + SeqSero2_version)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	35 return parser.parse_args()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	36
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	37 ### ed_SL_05282019: check paths of dependencies
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	38 check_dependencies = parse_args().check
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	39 dependencies = ['bwa','samtools','blastn','fastq-dump','spades.py','bedtools','SalmID.py']
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	40 if check_dependencies:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	41 for item in dependencies:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	42 ext_path = find_executable(item)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	43 if ext_path is not None:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	44 print ("Using "+item+" - "+ext_path)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	45 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	46 print ("ERROR: can not find "+item+" in PATH")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	47 sys.exit()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	48 ### end of --check
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	49
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	50 def reverse_complement(sequence):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	51 complement = {
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	52 'A': 'T',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	53 'C': 'G',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	54 'G': 'C',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	55 'T': 'A',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	56 'N': 'N',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	57 'M': 'K',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	58 'R': 'Y',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	59 'W': 'W',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	60 'S': 'S',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	61 'Y': 'R',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	62 'K': 'M',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	63 'V': 'B',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	64 'H': 'D',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	65 'D': 'H',
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	66 'B': 'V'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	67 }
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	68 return "".join(complement[base] for base in reversed(sequence))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	69
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	70
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	71 def createKmerDict_reads(list_of_strings, kmer):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	72 kmer_table = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	73 for string in list_of_strings:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	74 sequence = string.strip('\n')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	75 for i in range(len(sequence) - kmer + 1):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	76 new_mer = sequence[i:i + kmer].upper()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	77 new_mer_rc = reverse_complement(new_mer)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	78 if new_mer in kmer_table:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	79 kmer_table[new_mer.upper()] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	80 else:
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 if new_mer_rc in kmer_table:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	83 kmer_table[new_mer_rc.upper()] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	84 else:
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 return kmer_table
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	87
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	88
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	89 def multifasta_dict(multifasta):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	90 multifasta_list = [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	91 line.strip() for line in open(multifasta, 'r') if len(line.strip()) > 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	92 ]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	93 headers = [i for i in multifasta_list if i[0] == '>']
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	94 multifasta_dict = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	95 for h in headers:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	96 start = multifasta_list.index(h)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	97 for element in multifasta_list[start + 1:]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	98 if element[0] == '>':
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	99 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	100 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	101 if h[1:] in multifasta_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	102 multifasta_dict[h[1:]] += element
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	103 else:
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 return multifasta_dict
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	106
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	107
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	108 def multifasta_single_string(multifasta):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	109 multifasta_list = [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	110 line.strip() for line in open(multifasta, 'r')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	111 if (len(line.strip()) > 0) and (line.strip()[0] != '>')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	112 ]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	113 return ''.join(multifasta_list)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	114
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	115
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	116 def chunk_a_long_sequence(long_sequence, chunk_size=60):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	117 chunk_list = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	118 steps = len(long_sequence) // 60 #how many chunks
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	119 for i in range(steps):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	120 chunk_list.append(long_sequence[i * chunk_size:(i + 1) * chunk_size])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	121 chunk_list.append(long_sequence[steps * chunk_size:len(long_sequence)])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	122 return chunk_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	123
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	124
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	125 def target_multifasta_kmerizer(multifasta, k, kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	126 forward_length = 300 #if find the target, put forward 300 bases
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	127 reverse_length = 2200 #if find the target, put backward 2200 bases
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	128 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	129 target_mers = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	130 long_single_string = multifasta_single_string(multifasta)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	131 multifasta_list = chunk_a_long_sequence(long_single_string, chunk_size)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	132 unit_length = len(multifasta_list[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	133 forward_lines = int(forward_length / unit_length) + 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	134 reverse_lines = int(forward_length / unit_length) + 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	135 start_num = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	136 end_num = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	137 for i in range(len(multifasta_list)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	138 if i not in range(start_num, end_num): #avoid computational repetition
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	139 line = multifasta_list[i]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	140 start = int((len(line) - k) // 2)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	141 s1 = line[start:k + start]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	142 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	143 if i - forward_lines >= 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	144 start_num = i - forward_lines
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	145 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	146 start_num = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	147 if i + reverse_lines <= len(multifasta_list) - 1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	148 end_num = i + reverse_lines
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	149 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	150 end_num = len(multifasta_list) - 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	151 target_list = [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	152 x.strip() for x in multifasta_list[start_num:end_num]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	153 ]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	154 target_line = "".join(target_list)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	155 target_mers += [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	156 k1 for k1 in createKmerDict_reads([str(target_line)], k)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	157 ] ##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	158 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	159 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	160 return set(target_mers)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	161
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	162
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	163 def target_read_kmerizer(file, k, kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	164 i = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	165 n_reads = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	166 total_coverage = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	167 target_mers = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	168 if file.endswith(".gz"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	169 file_content = io.BufferedReader(gzip.open(file))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	170 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	171 file_content = open(file, "r").readlines()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	172 for line in file_content:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	173 start = int((len(line) - k) // 2)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	174 if i % 4 == 2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	175 if file.endswith(".gz"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	176 s1 = line[start:k + start].decode()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	177 line = line.decode()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	178 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	179 s1 = line[start:k + start]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	180 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	181 n_reads += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	182 total_coverage += len(line)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	183 target_mers += [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	184 k1 for k1 in createKmerDict_reads([str(line)], k)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	185 ] #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	186 i += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	187 if total_coverage >= 4000000:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	188 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	189 return set(target_mers)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	190
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	191
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	192 def minion_fasta_kmerizer(file, k, kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	193 i = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	194 n_reads = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	195 total_coverage = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	196 target_mers = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	197 for line in open(file):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	198 if i % 2 == 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	199 for kmer, rc_kmer in kmers(line.strip().upper(), k):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	200 if (kmer in kmerDict) or (rc_kmer in kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	201 if kmer in target_mers:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	202 target_mers[kmer] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	203 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	204 target_mers[kmer] = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	205 if rc_kmer in target_mers:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	206 target_mers[rc_kmer] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	207 else:
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 i += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	210 return set([h for h in target_mers])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	211
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	212
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	213 def minion_fastq_kmerizer(file, k, kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	214 i = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	215 n_reads = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	216 total_coverage = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	217 target_mers = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	218 for line in open(file):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	219 if i % 4 == 2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	220 for kmer, rc_kmer in kmers(line.strip().upper(), k):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	221 if (kmer in kmerDict) or (rc_kmer in kmerDict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	222 if kmer in target_mers:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	223 target_mers[kmer] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	224 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	225 target_mers[kmer] = 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	226 if rc_kmer in target_mers:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	227 target_mers[rc_kmer] += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	228 else:
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 i += 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	231 return set([h for h in target_mers])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	232
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	233
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	234 def multifasta_single_string2(multifasta):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	235 single_string = ''
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	236 with open(multifasta, 'r') as f:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	237 for line in f:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	238 if line.strip()[0] == '>':
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	239 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	240 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	241 single_string += line.strip()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	242 return single_string
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	243
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	244
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	245 def kmers(seq, k):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	246 rev_comp = reverse_complement(seq)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	247 for start in range(1, len(seq) - k + 1):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	248 yield seq[start:start + k], rev_comp[-(start + k):-start]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	249
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	250
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	251 def multifasta_to_kmers_dict(multifasta,k_size):#used to create database kmer set
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	252 multi_seq_dict = multifasta_dict(multifasta)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	253 lib_dict = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	254 for h in multi_seq_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	255 lib_dict[h] = set(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	256 [k for k in createKmerDict_reads([multi_seq_dict[h]], k_size)])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	257 return lib_dict
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	258
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	259
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	260 def Combine(b, c):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	261 fliC_combinations = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	262 fliC_combinations.append(",".join(c))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	263 temp_combinations = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	264 for i in range(len(b)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	265 for x in itertools.combinations(b, i + 1):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	266 temp_combinations.append(",".join(x))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	267 for x in temp_combinations:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	268 temp = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	269 for y in c:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	270 temp.append(y)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	271 temp.append(x)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	272 temp = ",".join(temp)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	273 temp = temp.split(",")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	274 temp.sort()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	275 temp = ",".join(temp)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	276 fliC_combinations.append(temp)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	277 return fliC_combinations
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	278
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	279
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	280 def seqsero_from_formula_to_serotypes(Otype, fliC, fljB, special_gene_list,subspecies):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	281 #like test_output_06012017.txt
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	282 #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	283 from Initial_Conditions import phase1,phase2,phaseO,sero,subs,remove_list,rename_dict
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	284 rename_dict_not_anymore=[rename_dict[x] for x in rename_dict]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	285 rename_dict_all=rename_dict_not_anymore+list(rename_dict) #used for decide whether to
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	286 seronames = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	287 seronames_none_subspecies=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	288 for i in range(len(phase1)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	289 fliC_combine = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	290 fljB_combine = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	291 if phaseO[i] == Otype: # no VII in KW, but it's there
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	292 ### for fliC, detect every possible combinations to avoid the effect of "["
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	293 if phase1[i].count("[") == 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	294 fliC_combine.append(phase1[i])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	295 elif phase1[i].count("[") >= 1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	296 c = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	297 b = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	298 if phase1[i][0] == "[" and phase1[i][-1] == "]" and phase1[i].count(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	299 "[") == 1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	300 content = phase1[i].replace("[", "").replace("]", "")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	301 fliC_combine.append(content)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	302 fliC_combine.append("-")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	303 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	304 for x in phase1[i].split(","):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	305 if "[" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	306 b.append(x.replace("[", "").replace("]", ""))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	307 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	308 c.append(x)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	309 fliC_combine = Combine(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	310 b, c
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	311 ) #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	312 ### end of fliC "[" detect
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	313 ### for fljB, detect every possible combinations to avoid the effect of "["
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	314 if phase2[i].count("[") == 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	315 fljB_combine.append(phase2[i])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	316 elif phase2[i].count("[") >= 1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	317 d = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	318 e = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	319 if phase2[i][0] == "[" and phase2[i][-1] == "]" and phase2[i].count(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	320 "[") == 1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	321 content = phase2[i].replace("[", "").replace("]", "")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	322 fljB_combine.append(content)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	323 fljB_combine.append("-")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	324 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	325 for x in phase2[i].split(","):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	326 if "[" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	327 d.append(x.replace("[", "").replace("]", ""))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	328 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	329 e.append(x)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	330 fljB_combine = Combine(d, e)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	331 ### end of fljB "[" detect
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	332 new_fliC = fliC.split(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	333 ","
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	334 ) #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	335 new_fliC.sort()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	336 new_fliC = ",".join(new_fliC)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	337 new_fljB = fljB.split(",")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	338 new_fljB.sort()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	339 new_fljB = ",".join(new_fljB)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	340 if (new_fliC in fliC_combine
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	341 or fliC in fliC_combine) and (new_fljB in fljB_combine
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	342 or fljB in fljB_combine):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	343 ######start, remove_list,rename_dict, added on 11/11/2018
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	344 if sero[i] not in remove_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	345 temp_sero=sero[i]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	346 if temp_sero in rename_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	347 temp_sero=rename_dict[temp_sero] #rename if in the rename list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	348 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	349 if subs[i] == subspecies:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	350 seronames.append(temp_sero)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	351 seronames_none_subspecies.append(temp_sero)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	352 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	353 pass
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 ######end, added on 11/11/2018
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	357 #analyze seronames
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	358 subspecies_pointer=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	359 if len(seronames) == 0 and len(seronames_none_subspecies)!=0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	360 seronames=seronames_none_subspecies
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	361 subspecies_pointer="1"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	362 if len(seronames) == 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	363 seronames = [
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	364 "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	365 ]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	366 star = ""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	367 star_line = ""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	368 if len(seronames) > 1: #there are two possible predictions for serotypes
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	369 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	370 #changed 04072019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	371 #star_line = "The predicted serotypes share the same general formula:\t" + Otype + ":" + fliC + ":" + fljB + "\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	372 if subspecies_pointer=="1" and len(seronames_none_subspecies)!=0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	373 star="*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	374 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	375 #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	376 if Otype=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	377 Otype="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	378 predict_form = Otype + ":" + fliC + ":" + fljB
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	379 predict_sero = (" or ").join(seronames)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	380 ###special test for Enteritidis
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	381 if predict_form == "9:g,m:-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	382 sdf = "-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	383 for x in special_gene_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	384 if x.startswith("sdf"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	385 sdf = "+"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	386 #star_line="Detected sdf gene, a marker to differentiate Gallinarum and Enteritidis"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	387 star_line=" sdf gene detected." # ed_SL_04152019: new output format
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	388 #predict_form = predict_form + " Sdf prediction:" + sdf
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	389 predict_form = predict_form #changed 04072019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	390 if sdf == "-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	391 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	392 #star_line="Didn't detected sdf gene, a marker to differentiate Gallinarum and Enteritidis"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	393 star_line=" sdf gene not detected." # ed_SL_04152019: new output format
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	394 #changed in 04072019, for new output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	395 #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	396 #predict_sero = "Gallinarum/Enteritidis" #04132019, for new output requirement
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	397 predict_sero = "Gallinarum or Enteritidis" # ed_SL_04152019: new output format
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	398 ###end of special test for Enteritidis
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	399 elif predict_form == "4:i:-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	400 predict_sero = "I 4,[5],12:i:-" # ed_SL_09242019: change serotype name
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	401 elif predict_form == "4:r:-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	402 predict_sero = "4:r:-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	403 elif predict_form == "4:b:-": # ed_SL_09272019: change for new output format
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	404 predict_sero = "N/A (4:b:-)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	405 #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	406 #predict_sero = "Newport"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	407 #star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	408 #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	409 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	410 if "N/A" in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	411 claim = ""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	412 #special test for Typhimurium
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	413 if "Typhimurium" in predict_sero or predict_form == "4:i:-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	414 normal = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	415 mutation = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	416 for x in special_gene_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	417 if "oafA-O-4_full" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	418 normal = float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	419 elif "oafA-O-4_5-" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	420 mutation = float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	421 if normal > mutation:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	422 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	423 elif normal < mutation:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	424 #predict_sero = predict_sero.strip() + "(O5-)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	425 predict_sero = predict_sero.strip() #diable special sero for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	426 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	427 #star_line = "Detected the deletion of O5-."
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	428 star_line = " Detected a deletion that causes O5- variant of Typhimurium." # ed_SL_04152019: new output format
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	429 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	430 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	431 #special test for Paratyphi B
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	432 if "Paratyphi B" in predict_sero or predict_form == "4:b:-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	433 normal = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	434 mutation = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	435 for x in special_gene_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	436 if "gntR-family-regulatory-protein_dt-positive" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	437 normal = float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	438 elif "gntR-family-regulatory-protein_dt-negative" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	439 mutation = float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	440 #print(normal,mutation)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	441 if normal > mutation:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	442 #predict_sero = predict_sero.strip() + "(dt+)" #diable special sero for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	443 predict_sero = predict_sero.strip()+' var. L(+) tartrate+' if "Paratyphi B" in predict_sero else predict_sero.strip() # ed_SL_04152019: new output format
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	444 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	445 #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	446 star_line = " The SNP that causes d-Tartrate nonfermentating phenotype of Paratyphi B was not detected. " # ed_SL_04152019: new output format
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	447 elif normal < mutation:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	448 #predict_sero = predict_sero.strip() + "(dt-)" #diable special sero for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	449 predict_sero = predict_sero.strip()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	450 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	451 #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	452 star_line = " Detected the SNP for d-Tartrate nonfermenting phenotype of Paratyphi B." # ed_SL_04152019: new output format
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	453 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	454 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	455 #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	456 star_line = " " ## ed_SL_05152019: do not report this situation.
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	457 #special test for O13,22 and O13,23
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	458 if Otype=="13":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	459 #ex_dir = os.path.dirname(os.path.realpath(__file__))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	460 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	461 f = open(ex_dir + '/special.pickle', 'rb')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	462 special = pickle.load(f)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	463 O22_O23=special['O22_O23']
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	464 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	465 O22_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	466 O23_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	467 for x in special_gene_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	468 if "O:22" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	469 O22_score = O22_score+float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	470 elif "O:23" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	471 O23_score = O23_score+float(special_gene_list[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	472 #print(O22_score,O23_score)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	473 for z in O22_O23[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	474 if predict_sero.split(" or ")[0] in z:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	475 if O22_score > O23_score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	476 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	477 #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	478 predict_sero = z[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	479 elif O22_score < O23_score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	480 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	481 #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	482 predict_sero = z[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	483 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	484 star = "*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	485 #star_line = "Fail to detect O22 and O23 differences." #diabled for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	486 if " or " in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	487 star_line = star_line + " The predicted serotypes share the same general formula:\t" + Otype + ":" + fliC + ":" + fljB + "\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	488 #special test for O6,8
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	489 #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	490 #for x in merge_O68_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	491 # if x in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	492 # predict_sero=x
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	493 # star=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	494 # star_line=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	495 #special test for Montevideo; most of them are monophasic
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	496 #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	497 #star="*"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	498 #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	499 return predict_form, predict_sero, star, star_line, claim
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	500 ### End of SeqSero Kmer part
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	501
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	502 ### Begin of SeqSero2 allele prediction and output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	503 def xml_parse_score_comparision_seqsero(xmlfile):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	504 #used to do seqsero xml analysis
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	505 from Bio.Blast import NCBIXML
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	506 handle=open(xmlfile)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	507 handle=NCBIXML.parse(handle)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	508 handle=list(handle)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	509 List=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	510 List_score=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	511 List_ids=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	512 List_query_region=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	513 for i in range(len(handle)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	514 if len(handle[i].alignments)>0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	515 for j in range(len(handle[i].alignments)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	516 score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	517 ids=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	518 cover_region=set() #fixed problem that repeated calculation leading percentage > 1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	519 List.append(handle[i].query.strip()+"___"+handle[i].alignments[j].hit_def)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	520 for z in range(len(handle[i].alignments[j].hsps)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	521 hsp=handle[i].alignments[j].hsps[z]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	522 temp=set(range(hsp.query_start,hsp.query_end))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	523 est_bit=(handle[i].ka_params[0]*hsp.score-math.log(handle[i].ka_params[1]))/(math.log(2))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	524 if len(cover_region)==0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	525 cover_region=cover_region\|temp
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	526 fraction=1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	527 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	528 fraction=1-len(cover_region&temp)/float(len(temp))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	529 cover_region=cover_region\|temp
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	530 if "last" in handle[i].query or "first" in handle[i].query:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	531 #score+=hsp.bits*fraction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	532 score+=est_bit*fraction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	533 ids+=float(hsp.identities)/handle[i].query_length*fraction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	534 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	535 #score+=hsp.bits*fraction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	536 score+=est_bit*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 List_score.append(score)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	539 List_ids.append(ids)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	540 List_query_region.append(cover_region)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	541 temp=zip(List,List_score,List_ids,List_query_region)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	542 Final_list=sorted(temp, key=lambda d:d[1], reverse = True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	543 return Final_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	544
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	545
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	546 def Uniq(L,sort_on_fre="none"): #return the uniq list and the count number
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	547 Old=L
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	548 L.sort()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	549 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	550 count=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	551 for j in range(len(L)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	552 y=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	553 for x in Old:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	554 if L[j]==x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	555 y+=1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	556 count.append(y)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	557 if sort_on_fre!="none":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	558 d=zip(*sorted(zip(count, L)))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	559 L=d[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	560 count=d[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	561 return (L,count)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	562
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	563 def judge_fliC_or_fljB_from_head_tail_for_one_contig(nodes_vs_score_list):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	564 #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	565 #this is mainly used for
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	566 a=nodes_vs_score_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	567 fliC_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	568 fljB_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	569 for z in a:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	570 if "fliC" in z[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	571 fliC_score+=z[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	572 elif "fljB" in z[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	573 fljB_score+=z[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	574 if fliC_score>=fljB_score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	575 role="fliC"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	576 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	577 role="fljB"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	578 return (role,abs(fliC_score-fljB_score))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	579
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	580 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	581 #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	582 #also used when no head or tail got blasted score for the contig
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	583 role=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	584 for z in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	585 if node_name in z[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	586 role=z[0].split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	587 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	588 return role
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	589
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	590 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	591 #nodes_list is the c created by c,d=Uniq(nodes) in below function
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	592 first_target=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	593 role_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	594 for x in nodes_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	595 a=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	596 role=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	597 for y in tail_head_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	598 if x in y[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	599 a.append(y)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	600 if len(a)==4:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	601 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	602 if diff<20:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	603 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	604 elif len(a)==3:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	605 ###however, if the one with highest score is the fewer one, compare their accumulation score
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	606 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	607 if diff<20:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	608 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	609 ###end of above score comparison
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	610 elif len(a)==2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	611 #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	612 temp=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	613 for z in a:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	614 temp.append(z[0].split("_")[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	615 m,n=Uniq(temp)#should only have one choice, but weird situation might occur too
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	616 if len(m)==1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	617 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	618 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	619 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	620 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	621 if diff<20:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	622 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	623 ###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	624 elif len(a)==1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	625 #that one
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	626 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	627 if diff<20:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	628 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	629 #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	630 else:#a==0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	631 #use Final_list_passed best match
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	632 for z in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	633 if x in z[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	634 role=z[0].split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	635 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	636 #print x,role,len(a)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	637 role_list.append((role,x))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	638 if len(role_list)==2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	639 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	640 #just use score to do a final test
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	641 role_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	642 for x in nodes_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	643 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	644 role_list.append((role,x))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	645 return role_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	646
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	647 def decide_contig_roles_for_H_antigen(Final_list,Final_list_passed):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	648 #used to decide which contig is FliC and which one is fljB
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	649 contigs=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	650 nodes=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	651 for x in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	652 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	653 nodes.append(x[0].split("___")[1].strip())
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	654 c,d=Uniq(nodes)#c is node_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	655 #print c
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	656 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	657 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	658 return roles
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	659
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	660 def decide_O_type_and_get_special_genes(Final_list,Final_list_passed):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	661 #decide O based on Final_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	662 O_choice="?"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	663 O_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	664 special_genes={}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	665 nodes=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	666 for x in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	667 if x[0].startswith("O-"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	668 nodes.append(x[0].split("___")[1].strip())
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	669 elif not x[0].startswith("fl"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	670 special_genes[x[0]]=x[2]#08172018, x[2] changed from x[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	671 #print "special_genes:",special_genes
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	672 c,d=Uniq(nodes)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	673 #print "potential O antigen contig",c
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	674 final_O=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	675 O_nodes_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	676 for x in c:#c is the list for contigs
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	677 temp=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	678 for y in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	679 if x in y[0] and y[0].startswith("O-"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	680 final_O.append(y)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	681 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	682 ### O contig has the problem of two genes on same contig, so do additional test
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	683 potenial_new_gene=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	684 for x in final_O:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	685 pointer=0 #for genes merged or not
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	686 #not consider O-1,3,19_not_in_3,10, too short compared with others
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	687 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	688 pointer=x[0].split("___")[1].strip()#store the contig name
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	689 print(pointer)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	690 if pointer!=0:#it has potential merge event
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	691 for y in Final_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	692 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	693 potenial_new_gene=y
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	694 #print(potenial_new_gene)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	695 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	696 if potenial_new_gene!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	697 print("two differnt genes in same contig, fix it for O antigen")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	698 print(potenial_new_gene[:3])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	699 pointer=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	700 for y in final_O:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	701 if y[0].split("___")[-1]==potenial_new_gene[0].split("___")[-1]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	702 pointer=1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	703 if pointer!=0: #changed to consider two genes in same contig
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	704 final_O.append(potenial_new_gene)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	705 ### end of the two genes on same contig test
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	706 final_O=sorted(final_O,key=lambda x: x[2], reverse=True)#sorted
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	707 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	708 #print "$$$No Otype, due to no hit"#may need to be changed
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	709 O_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	710 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	711 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	712 O_list=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	713 O_list_less_contamination=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	714 for x in final_O:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	715 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	716 O_list.append(x[0].split("__")[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	717 O_nodes_list.append(x[0].split("___")[1])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	718 if float(x[0].split("_cov_")[-1].split("_")[0])>highest_O_coverage*0.15:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	719 O_list_less_contamination.append(x[0].split("__")[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	720 ### special test for O9,46 and O3,10 family
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	721 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	722 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	723 O_choice="O-9,46"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	724 #print "$$$Most possilble Otype: O-9,46"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	725 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	726 O_choice="O-9,46,27"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	727 #print "$$$Most possilble Otype: O-9,46,27"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	728 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	729 O_choice="O-9"#next, detect O9 vs O2?
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	730 O2=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	731 O9=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	732 for z in special_genes:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	733 if "tyr-O-9" in z:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	734 O9=special_genes[z]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	735 elif "tyr-O-2" in z:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	736 O2=special_genes[z]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	737 if O2>O9:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	738 O_choice="O-2"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	739 elif O2<O9:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	740 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	741 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	742 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	743 #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	744 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	745 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	746 O_choice="O-3,10"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	747 #print "$$$Most possilble Otype: O-3,10 (contain O-3,10_not_in_1,3,19)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	748 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	749 O_choice="O-1,3,19"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	750 #print "$$$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	751 ### end of special test for O9,46 and O3,10 family
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	752 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	753 try:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	754 max_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	755 for x in final_O:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	756 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	757 max_score=x[2]#change from x[-1] to x[2],08172018
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	758 O_choice=x[0].split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	759 if O_choice=="O-1,3,19":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	760 O_choice=final_O[1][0].split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	761 #print "$$$Most possilble Otype: ",O_choice
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	762 except:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	763 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	764 #print "$$$No suitable Otype, or failure of mapping (please check the quality of raw reads)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	765 #print "O:",O_choice,O_nodes_list
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	766 Otypes=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	767 for x in O_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	768 if x!="O-1,3,19_not_in_3,10":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	769 if "O-9,46_" not in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	770 Otypes.append(x.split("_")[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	771 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	772 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	773 #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	774 Otypes_uniq,Otypes_fre=Uniq(Otypes)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	775 contamination_O=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	776 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	777 if len(Otypes_uniq)>2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	778 contamination_O="potential contamination from O antigen signals"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	779 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	780 if len(Otypes_uniq)>1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	781 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	782 contamination_O=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	783 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	784 contamination_O=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	785 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	786 contamination_O="potential contamination from O antigen signals"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	787 return O_choice,O_nodes_list,special_genes,final_O,contamination_O,Otypes_uniq
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	788 ### End of SeqSero2 allele prediction and output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	789
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	790 def get_input_files(make_dir,input_file,data_type,dirpath):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	791 #tell input files from datatype
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	792 #"<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	793 for_fq=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	794 rev_fq=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	795 os.chdir(make_dir)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	796 if data_type=="1":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	797 input_file=input_file[0].split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	798 if input_file.endswith(".sra"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	799 subprocess.check_call("fastq-dump --split-files "+input_file,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	800 for_fq=input_file.replace(".sra","_1.fastq")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	801 rev_fq=input_file.replace(".sra","_2.fastq")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	802 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	803 core_id=input_file.split(".fastq")[0].split(".fq")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	804 for_fq=core_id+"_1.fastq"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	805 rev_fq=core_id+"_2.fastq"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	806 if input_file.endswith(".gz"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	807 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	808 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	809 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	810 elif data_type=="2":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	811 for_fq=input_file[0].split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	812 rev_fq=input_file[1].split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	813 elif data_type=="3":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	814 input_file=input_file[0].split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	815 if input_file.endswith(".sra"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	816 subprocess.check_call("fastq-dump --split-files "+input_file,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	817 for_fq=input_file.replace(".sra","_1.fastq")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	818 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	819 for_fq=input_file
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	820 elif data_type in ["4","5","6"]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	821 for_fq=input_file[0].split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	822 os.chdir("..")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	823 return for_fq,rev_fq
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	824
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	825 def predict_O_and_H_types(Final_list,Final_list_passed,new_fasta):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	826 #get O and H types from Final_list from blast parsing; allele mode
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	827 from Bio import SeqIO
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	828 fliC_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	829 fljB_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	830 fliC_contig="NA"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	831 fljB_contig="NA"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	832 fliC_region=set([0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	833 fljB_region=set([0,])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	834 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	835 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	836 O_choice="-"#no need to decide O contig for now, should be only one
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	837 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	838 O_choice=O_choice.split("-")[-1].strip()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	839 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	840 O_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	841 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	842 #add alignment locations, used for further selection, 03312019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	843 for i in range(len(H_contig_roles)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	844 x=H_contig_roles[i]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	845 for y in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	846 if x[1] in y[0] and y[0].startswith(x[0]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	847 H_contig_roles[i]+=H_contig_roles[i]+(y[-1],)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	848 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	849 log_file=open("SeqSero_log.txt","a")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	850 extract_file=open("Extracted_antigen_alleles.fasta","a")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	851 handle_fasta=list(SeqIO.parse(new_fasta,"fasta"))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	852
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	853 #print("O_contigs:")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	854 log_file.write("O_contigs:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	855 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	856 extract_file.write("#O_contigs:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	857 for x in O_nodes_roles:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	858 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	859 #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	860 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	861 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	862 seqs=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	863 for z in handle_fasta:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	864 if x[0].split("___")[-1]==z.description:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	865 seqs=str(z.seq)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	866 extract_file.write(title+seqs+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	867 if len(H_contig_roles)!=0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	868 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	869 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	870 highest_H_coverage=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	871 for x in H_contig_roles:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	872 #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	873 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	874 fliC_contig=x[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	875 fliC_length=len(x[-1])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	876 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	877 fljB_contig=x[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	878 fljB_length=len(x[-1])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	879 for x in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	880 if fliC_choice=="-" and "fliC_" in x[0] and fliC_contig in x[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	881 fliC_choice=x[0].split("_")[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	882 elif fljB_choice=="-" and "fljB_" in x[0] and fljB_contig in x[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	883 fljB_choice=x[0].split("_")[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	884 elif fliC_choice!="-" and fljB_choice!="-":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	885 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	886 #now remove contigs not in middle core part
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	887 first_allele="NA"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	888 first_allele_percentage=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	889 for x in Final_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	890 if x[0].startswith("fliC") or x[0].startswith("fljB"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	891 first_allele=x[0].split("__")[0] #used to filter those un-middle contigs
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	892 first_allele_percentage=x[2]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	893 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	894 additional_contigs=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	895 for x in Final_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	896 if first_allele in x[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	897 if (fliC_contig == x[0].split("___")[-1]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	898 fliC_region=x[3]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	899 elif fljB_contig!="NA" and (fljB_contig == x[0].split("___")[-1]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	900 fljB_region=x[3]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	901 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	902 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	903 additional_contigs.append(x)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	904 #else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	905 #print x[:3]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	906 #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	907 if first_allele_percentage>0.9:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	908 if len(fliC_region)>len(fljB_region) and (max(fljB_region)-min(fljB_region))>1000:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	909 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	910 elif len(fliC_region)<len(fljB_region) and (max(fliC_region)-min(fliC_region))>1000:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	911 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	912 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	913 target_region=set()#doesn't do anything
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	914 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	915 target_region=set()#doesn't do anything
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	916 #print(target_region)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	917 #print(additional_contigs)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	918 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	919 target_region=target_region2\|target_region
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	920 for x in additional_contigs:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	921 removal=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	922 contig_length=int(x[0].split("___")[1].split("length_")[-1].split("_")[0])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	923 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	924 removal=1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	925 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	926 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	927 removal=1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	928 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	929 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	930 if removal==1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	931 for y in H_contig_roles:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	932 if y[1] in x[0]:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	933 H_contig_roles.remove(y)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	934 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	935 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	936 #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	937 #end of removing none-middle contigs
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	938 #print("H_contigs:")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	939 log_file.write("H_contigs:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	940 extract_file.write("#H_contigs:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	941 H_contig_stat=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	942 H1_cont_stat={}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	943 H2_cont_stat={}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	944 for i in range(len(H_contig_roles)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	945 x=H_contig_roles[i]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	946 a=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	947 for y in Final_list_passed:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	948 if x[1] in y[0] and y[0].startswith(x[0]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	949 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	950 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	951 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	952 #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	953 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	954 H_contig_roles[i]="can't decide fliC or fljB, may be third phase"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	955 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	956 seqs=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	957 for z in handle_fasta:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	958 if x[1]==z.description:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	959 seqs=str(z.seq)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	960 extract_file.write(title+seqs+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	961 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	962 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	963 #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	964 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	965 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	966 seqs=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	967 for z in handle_fasta:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	968 if x[1]==z.description:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	969 seqs=str(z.seq)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	970 extract_file.write(title+seqs+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	971 if x[0]=="fliC":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	972 if y[0].split("_")[1] not in H1_cont_stat:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	973 H1_cont_stat[y[0].split("_")[1]]=y[2]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	974 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	975 H1_cont_stat[y[0].split("_")[1]]+=y[2]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	976 if x[0]=="fljB":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	977 if y[0].split("_")[1] not in H2_cont_stat:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	978 H2_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 H2_cont_stat[y[0].split("_")[1]]+=y[2]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	981 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	982 #detect contaminations
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	983 #print(H1_cont_stat)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	984 #print(H2_cont_stat)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	985 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	986 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	987 contamination_H=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	988 if len(H1_cont_stat_list)>1 or len(H2_cont_stat_list)>1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	989 contamination_H="potential contamination from H antigen signals"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	990 elif len(H2_cont_stat_list)==1 and fljB_contig=="NA":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	991 contamination_H="potential contamination from H antigen signals, uncommon weak fljB signals detected"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	992 #get additional antigens
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	993 """
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	994 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	995 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	996 O_choice="O-9,46"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	997 #print "$$$Most possilble Otype: O-9,46"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	998 elif "O-9,46,27_partial_wzy" in O_list:#and float(O94627)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	999 O_choice="O-9,46,27"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1000 #print "$$$Most possilble Otype: O-9,46,27"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1001 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	1002 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	1003 O_choice="O-3,10"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1004 #print "$$$Most possilble Otype: O-3,10 (contain O-3,10_not_in_1,3,19)"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1005 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1006 O_choice="O-1,3,19"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1007 #print "$$$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	1008 ### end of special test for O9,46 and O3,10 family
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1009
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1010 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	1011 if len(Otypes_uniq)>2:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1012 contamination_O="potential contamination from O antigen signals"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1013 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1014 if len(Otypes_uniq)>1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1015 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	1016 contamination_O=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1017 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	1018 contamination_O=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1019 """
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1020 additonal_antigents=[]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1021 #print(contamination_O)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1022 #print(contamination_H)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1023 log_file.write(contamination_O+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1024 log_file.write(contamination_H+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1025 log_file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1026 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	1027
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1028 def get_input_K(input_file,lib_dict,data_type,k_size):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1029 #kmer mode; get input_Ks from dict and data_type
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1030 kmers = []
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1031 for h in lib_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1032 kmers += lib_dict[h]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1033 if data_type == '4':
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1034 input_Ks = target_multifasta_kmerizer(input_file, k_size, set(kmers))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1035 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	1036 input_Ks = target_read_kmerizer(input_file, k_size, set(kmers))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1037 elif data_type == '5':#minion_2d_fasta
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1038 input_Ks = minion_fasta_kmerizer(input_file, k_size, set(kmers))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1039 if data_type == '6':#minion_2d_fastq
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1040 input_Ks = minion_fastq_kmerizer(input_file, k_size, set(kmers))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1041 return input_Ks
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1042
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1043 def get_kmer_dict(lib_dict,input_Ks):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1044 #kmer mode; get predicted types
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1045 O_dict = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1046 H_dict = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1047 Special_dict = {}
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1048 for h in lib_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1049 score = (len(lib_dict[h] & input_Ks) / len(lib_dict[h])) * 100
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1050 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	1051 if h.startswith('O-') and score > 25:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1052 O_dict[h] = score
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1053 if h.startswith('fl') and score > 40:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1054 H_dict[h] = score
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1055 if (h[:2] != 'fl') and (h[:2] != 'O-'):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1056 Special_dict[h] = score
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1057 return O_dict,H_dict,Special_dict
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1058
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1059 def call_O_and_H_type(O_dict,H_dict,Special_dict,make_dir):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1060 log_file=open("SeqSero_log.txt","a")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1061 log_file.write("O_scores:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1062 #call O:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1063 highest_O = '-'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1064 if len(O_dict) == 0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1065 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1066 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1067 for x in O_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1068 log_file.write(x+"\t"+str(O_dict[x])+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1069 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	1070 if 'O-9,46_wzy__1191' in O_dict: # and float(O946_wzy)/float(num_1) > 0.1
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1071 highest_O = "O-9,46"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1072 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	1073 highest_O = "O-9,46,27"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1074 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1075 highest_O = "O-9" # next, detect O9 vs O2?
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1076 O2 = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1077 O9 = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1078 for z in Special_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1079 if "tyr-O-9" in z:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1080 O9 = float(Special_dict[z])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1081 if "tyr-O-2" in z:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1082 O2 = float(Special_dict[z])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1083 if O2 > O9:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1084 highest_O = "O-2"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1085 elif ("O-3,10_wzx__1539" in O_dict) and (
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1086 "O-9,46_wzy__1191" in O_dict
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1087 ): # 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	1088 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	1089 highest_O = "O-3,10"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1090 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1091 highest_O = "O-1,3,19"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1092 ### end of special test for O9,46 and O3,10 family
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1093 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1094 try:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1095 max_score = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1096 for x in O_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1097 if float(O_dict[x]) >= max_score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1098 max_score = float(O_dict[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1099 highest_O = x.split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1100 if highest_O == "O-1,3,19":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1101 highest_O = '-'
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 x == 'O-1,3,19_not_in_3,10__130':
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1105 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1106 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1107 if float(O_dict[x]) >= max_score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1108 max_score = float(O_dict[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1109 highest_O = x.split("_")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1110 except:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1111 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1112 #call_fliC:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1113 if len(H_dict)!=0:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1114 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	1115 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1116 highest_H_score_both_BC=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1117 highest_fliC = '-'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1118 highest_fliC_raw = '-'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1119 highest_Score = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1120 log_file.write("\nH_scores:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1121 for s in H_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1122 log_file.write(s+"\t"+str(H_dict[s])+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1123 if s.startswith('fliC'):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1124 if float(H_dict[s]) > highest_Score:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1125 highest_fliC = s.split('_')[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1126 highest_fliC_raw = s
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1127 highest_Score = float(H_dict[s])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1128 #call_fljB
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1129 highest_fljB = '-'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1130 highest_fljB_raw = '-'
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1131 highest_Score = 0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1132 for s in H_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1133 if s.startswith('fljB'):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1134 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	1135 highest_fljB = s.split('_')[1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1136 highest_fljB_raw = s
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1137 highest_Score = float(H_dict[s])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1138 log_file.write("\nSpecial_scores:\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1139 for s in Special_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1140 log_file.write(s+"\t"+str(Special_dict[s])+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1141 log_file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1142 return highest_O,highest_fliC,highest_fljB
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1143
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1144 def get_temp_file_names(for_fq,rev_fq):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1145 #seqsero2 -a; get temp file names
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1146 sam=for_fq+".sam"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1147 bam=for_fq+".bam"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1148 sorted_bam=for_fq+"_sorted.bam"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1149 mapped_fq1=for_fq+"_mapped.fq"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1150 mapped_fq2=rev_fq+"_mapped.fq"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1151 combined_fq=for_fq+"_combined.fq"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1152 for_sai=for_fq+".sai"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1153 rev_sai=rev_fq+".sai"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1154 return sam,bam,sorted_bam,mapped_fq1,mapped_fq2,combined_fq,for_sai,rev_sai
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1155
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1156 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	1157 #seqsero2 -a; do mapping and sort
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1158 print("building database...")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1159 subprocess.check_call("bwa index "+database+ " 2>> data_log.txt",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1160 print("mapping...")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1161 if mapping_mode=="mem":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1162 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	1163 elif mapping_mode=="sam":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1164 if fnameB!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1165 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	1166 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	1167 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	1168 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1169 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	1170 subprocess.check_call("bwa samse "+database+" "+for_sai+" "+for_fq+" > "+sam)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1171 subprocess.check_call("samtools view -@ "+threads+" -F 4 -Sh "+sam+" > "+bam,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1172 ### check the version of samtools then use differnt commands
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1173 samtools_version=subprocess.Popen(["samtools"],stdout=subprocess.PIPE,stderr=subprocess.PIPE)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1174 out, err = samtools_version.communicate()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1175 version = str(err).split("ersion:")[1].strip().split(" ")[0].strip()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1176 print("check samtools version:",version)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1177 ### end of samtools version check and its analysis
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1178 if LooseVersion(version)<=LooseVersion("1.2"):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1179 subprocess.check_call("samtools sort -@ "+threads+" -n "+bam+" "+fnameA+"_sorted",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1180 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1181 subprocess.check_call("samtools sort -@ "+threads+" -n "+bam+" >"+sorted_bam,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1182
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1183 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	1184 #seqsero2 -a; extract, assembly and blast
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1185 subprocess.check_call("bamToFastq -i "+sorted_bam+" -fq "+combined_fq,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1186 #print("fnameA:",fnameA)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1187 #print("fnameB:",fnameB)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1188 if fnameB!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1189 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	1190 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1191 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1192 outdir=current_time+"_temp"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1193 print("assembling...")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1194 if int(threads)>4:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1195 t="4"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1196 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1197 t=threads
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1198 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	1199 if fnameB!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1200 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	1201 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1202 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	1203 #new_fasta=fnameA+"_"+database+"_"+mapping_mode+".fasta"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1204 new_fasta=fnameA+"_"+database.split('/')[-1]+"_"+mapping_mode+".fasta" # ed_SL_09152019: change path to databse for packaging
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1205 subprocess.check_call("mv "+outdir+"/contigs.fasta "+new_fasta+ " 2> /dev/null",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1206 #os.system("mv "+outdir+"/scaffolds.fasta "+new_fasta+ " 2> /dev/null") contigs.fasta
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1207 subprocess.check_call("rm -rf "+outdir+ " 2> /dev/null",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1208 print("blasting...","\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1209 xmlfile="blasted_output.xml"#fnameA+"-extracted_vs_"+database+"_"+mapping_mode+".xml"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1210 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	1211 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	1212 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1213 xmlfile="NA"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1214 return xmlfile,new_fasta
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1215
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1216 def judge_subspecies(fnameA,dirpath):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1217 #seqsero2 -a; judge subspecies on just forward raw reads fastq
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1218 salmID_output=subprocess.Popen("python " + dirpath + "/SalmID.py -i "+fnameA,shell=True,stdout=subprocess.PIPE,stderr=subprocess.PIPE)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1219 out, err = salmID_output.communicate()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1220 out=out.decode("utf-8")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1221 file=open("data_log.txt","a")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1222 file.write(out)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1223 file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1224 salm_species_scores=out.split("\n")[1].split("\t")[6:]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1225 salm_species_results=out.split("\n")[0].split("\t")[6:]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1226 max_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1227 max_score_index=1 #default is 1, means "I"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1228 for i in range(len(salm_species_scores)):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1229 if max_score<float(salm_species_scores[i]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1230 max_score=float(salm_species_scores[i])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1231 max_score_index=i
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1232 prediction=salm_species_results[max_score_index].split(".")[1].strip().split(" ")[0]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1233 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	1234 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	1235 if max_score<10:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1236 prediction="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1237 return prediction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1238
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1239 def judge_subspecies_Kmer(Special_dict):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1240 #seqsero2 -k;
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1241 max_score=0
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1242 prediction="-" #default should be I
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1243 for x in Special_dict:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1244 if "mer" in x:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1245 if max_score<float(Special_dict[x]):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1246 max_score=float(Special_dict[x])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1247 prediction=x.split("_")[-1].strip()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1248 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	1249 prediction="bongori"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1250 break
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1251 return prediction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1252
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1253 def main():
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1254 #combine SeqSeroK and SeqSero2, also with SalmID
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1255 args = parse_args()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1256 input_file = args.i
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1257 data_type = args.t
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1258 analysis_mode = args.m
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1259 mapping_mode=args.b
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1260 threads=args.p
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1261 make_dir=args.d
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1262 clean_mode=args.c
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1263 k_size=27 #will change for bug fixing
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1264 #database="H_and_O_and_specific_genes.fasta"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1265 dirpath = os.path.abspath(os.path.dirname(os.path.realpath(__file__)))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1266 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	1267 #database=ex_dir+"/seqsero2_db/H_and_O_and_specific_genes.fasta" # ed_SL_09152019: change path to database for packaging
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1268 database="H_and_O_and_specific_genes.fasta"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1269 note="Note:"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1270 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	1271 if len(sys.argv)==1:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1272 subprocess.check_call(dirpath+"/SeqSero2_package.py -h",shell=True)#change name of python file
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1273 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1274 request_id = time.strftime("%m_%d_%Y_%H_%M_%S", time.localtime())
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1275 request_id += str(random.randint(1, 10000000))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1276 if make_dir is None:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1277 make_dir="SeqSero_result_"+request_id
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1278 if os.path.isdir(make_dir):
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1279 pass
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1280 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1281 subprocess.check_call(["mkdir",make_dir])
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1282 #subprocess.check_call("cp "+dirpath+"/"+database+" "+" ".join(input_file)+" "+make_dir,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1283 #subprocess.check_call("ln -sr "+dirpath+"/"+database+" "+" ".join(input_file)+" "+make_dir,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1284 #subprocess.check_call("ln -f -s "+database+" "+" ".join(input_file)+" "+make_dir,shell=True) # ed_SL_09152019: change path to database for packaging
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1285 subprocess.check_call("ln -f -s "+dirpath+"/seqsero2_db/"+database+" "+" ".join(input_file)+" "+make_dir,shell=True) ### ed_SL_05282019: 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	1286 ############################begin the real analysis
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1287 if analysis_mode=="a":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1288 if data_type in ["1","2","3"]:#use allele mode
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1289 for_fq,rev_fq=get_input_files(make_dir,input_file,data_type,dirpath)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1290 os.chdir(make_dir)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1291 ###add a function to tell input files
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1292 fnameA=for_fq.split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1293 fnameB=rev_fq.split("/")[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1294 current_time=time.strftime("%Y_%m_%d_%H_%M_%S", time.localtime())
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1295 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	1296 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	1297 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	1298 if xmlfile=="NA":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1299 O_choice,fliC_choice,fljB_choice,special_gene_list,contamination_O,contamination_H=("-","-","-",[],"","")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1300 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1301 Final_list=xml_parse_score_comparision_seqsero(xmlfile) #analyze xml and get parsed results
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1302 file=open("data_log.txt","a")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1303 for x in Final_list:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1304 file.write("\t".join(str(y) for y in x)+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1305 file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1306 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	1307 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
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1308 subspecies=judge_subspecies(fnameA,dirpath) #predict subspecies
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1309 ###output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1310 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	1311 claim="" #04132019, disable claim for new report requirement
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1312 contamination_report=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1313 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	1314 if contamination_O!="" and contamination_H=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1315 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	1316 elif contamination_O=="" and contamination_H!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1317 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	1318 elif contamination_O!="" and contamination_H!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1319 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	1320 if contamination_report!="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1321 #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	1322 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	1323 #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	1324 ## ed_SL_09272019: change for new output format
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1325 #if contamination_report+star_line+claim=="": #0413, new output style
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1326 # note=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1327 #else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1328 # note="Note:"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1329 if clean_mode:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1330 subprocess.check_call("rm -rf ../"+make_dir,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1331 make_dir="none-output-directory due to '-c' flag"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1332 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1333 new_file=open("SeqSero_result.txt","w")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1334 if O_choice=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1335 O_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1336 if "N/A" not in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1337 new_file.write("Output_directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1338 "Input files:\t"+"\t".join(input_file)+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1339 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1340 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1341 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1342 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1343 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1344 "Predicted serotype:\t"+predict_sero+"\n"+ # ed_SL_04152019: change serotype(s) to serotype
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1345 note+contamination_report+star_line+claim+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1346 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1347 #star_line=star_line.strip()+"\tNone such antigenic formula in KW.\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1348 star_line="" #04132019, for new output requirement, diable star_line if "NA" in output
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1349 new_file.write("Output_directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1350 "Input files:\t"+"\t".join(input_file)+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1351 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1352 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1353 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1354 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1355 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1356 "Predicted serotype:\t"+predict_form+"\n"+ # ed_SL_09242019: add serotype output for "N/A" prediction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1357 note+NA_note+contamination_report+star_line+claim+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1358 new_file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1359 print("\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1360 #subprocess.check_call("cat Seqsero_result.txt",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1361 #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	1362 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	1363 if "N/A" not in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1364 #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	1365 print("Output_directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1366 "Input files:\t"+"\t".join(input_file)+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1367 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1368 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1369 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1370 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1371 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1372 "Predicted serotype:\t"+predict_sero+"\n"+ # ed_SL_04152019: change serotype(s) to serotype
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1373 note+contamination_report+star_line+claim+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1374 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1375 print("Output_directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1376 "Input files:\t"+"\t".join(input_file)+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1377 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1378 "H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1379 "H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1380 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1381 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1382 "Predicted serotype:\t"+predict_form+"\n"+ # ed_SL_09242019: add serotype output for "N/A" prediction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1383 note+NA_note+contamination_report+star_line+claim+"\n")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1384 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1385 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	1386 elif analysis_mode=="k":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1387 #ex_dir = os.path.dirname(os.path.realpath(__file__))
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1388 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	1389 #output_mode = args.mode
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1390 for_fq,rev_fq=get_input_files(make_dir,input_file,data_type,dirpath)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1391 input_file = for_fq #-k will just use forward because not all reads were used
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1392 os.chdir(make_dir)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1393 f = open(dirpath + '/seqsero2_db/antigens.pickle', 'rb')
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1394 lib_dict = pickle.load(f)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1395 f.close
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1396 input_Ks=get_input_K(input_file,lib_dict,data_type,k_size)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1397 O_dict,H_dict,Special_dict=get_kmer_dict(lib_dict,input_Ks)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1398 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	1399 subspecies=judge_subspecies_Kmer(Special_dict)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1400 if subspecies=="IIb" or subspecies=="IIa":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1401 subspecies="II"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1402 predict_form,predict_sero,star,star_line,claim = seqsero_from_formula_to_serotypes(
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1403 highest_O.split('-')[1], highest_fliC, highest_fljB, Special_dict,subspecies)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1404 claim="" #no claim any more based on new output requirement
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1405 ## ed_SL_09272019: change for new output format
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1406 #if star_line+claim=="": #0413, new output style
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1407 # note=""
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1408 #else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1409 # note="Note:"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1410 if clean_mode:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1411 subprocess.check_call("rm -rf ../"+make_dir,shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1412 make_dir="none-output-directory due to '-c' flag"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1413 ### ed_SL_05282019, fix the assignment issue of variable 'O_choice' using "-m k -c"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1414 if highest_O.split('-')[-1]=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1415 O_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1416 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1417 O_choice=highest_O.split('-')[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1418 ###
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1419 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1420 if highest_O.split('-')[-1]=="":
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1421 O_choice="-"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1422 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1423 O_choice=highest_O.split('-')[-1]
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1424 #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	1425 new_file=open("SeqSero_result.txt","w")
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1426 #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	1427 if "N/A" not in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1428 new_file.write("Output_directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1429 "Input files:\t"+input_file+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1430 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1431 "H1 antigen prediction(fliC):\t"+highest_fliC+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1432 "H2 antigen prediction(fljB):\t"+highest_fljB+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1433 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1434 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1435 "Predicted serotype:\t"+predict_sero+"\n"+ # ed_SL_04152019: change serotype(s) to serotype
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1436 note+star_line+claim+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1437 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1438 #star_line=star_line.strip()+"\tNone such antigenic formula in KW.\n"
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1439 star_line = "" #changed for new output requirement, 04132019
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1440 new_file.write("Output_directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1441 "Input files:\t"+input_file+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1442 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1443 "H1 antigen prediction(fliC):\t"+highest_fliC+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1444 "H2 antigen prediction(fljB):\t"+highest_fljB+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1445 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1446 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1447 "Predicted serotype:\t"+predict_form+"\n"+ # ed_SL_09242019: add serotype output for "N/A" prediction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1448 note+NA_note+star_line+claim+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1449 new_file.close()
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1450 subprocess.call("rm .fasta .fastq .gz .fq temp.txt .sra 2> /dev/null",shell=True)
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1451 if "N/A" not in predict_sero:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1452 print("Output_directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1453 "Input files:\t"+input_file+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1454 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1455 "H1 antigen prediction(fliC):\t"+highest_fliC+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1456 "H2 antigen prediction(fljB):\t"+highest_fljB+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1457 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1458 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1459 "Predicted serotype:\t"+predict_sero+"\n"+ # ed_SL_04152019: change serotype(s) to serotype
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1460 note+star_line+claim+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1461 else:
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1462 print("Output_directory:\t"+make_dir+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1463 "Input files:\t"+input_file+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1464 "O antigen prediction:\t"+O_choice+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1465 "H1 antigen prediction(fliC):\t"+highest_fliC+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1466 "H2 antigen prediction(fljB):\t"+highest_fljB+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1467 "Predicted subspecies:\t"+subspecies+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1468 "Predicted antigenic profile:\t"+predict_form+"\n"+
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1469 "Predicted serotype:\t"+predict_form+"\n"+ # ed_SL_09242019: add serotype output for "N/A" prediction
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1470 note+NA_note+star_line+claim+"\n")#+##
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1471
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1472 if __name__ == '__main__':
fc22ec8e924e planemo upload commit 6b0a9d0f0ef4bdb0c2e2c54070b510ff28125f7a cstrittmatter parents: diff changeset	1473 main()

Mercurial > repos > cstrittmatter > ss2v110

annotate SeqSero2_package.py @ 9:43f6b7f6ebb3 draft