test_eurl_vtec_wgs_pt: scripts/ReMatCh/rematch.py annotate

annotate scripts/ReMatCh/rematch.py @ 3:0cbed1c0a762 draft default tip

planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8

author	cstrittmatter
date	Tue, 28 Jan 2020 10:42:31 -0500
parents	965517909457
children

rev	line source
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	1 #!/usr/bin/env python3
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	2
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	3 # -- coding: utf-8 --
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	4
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	5 """
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	6 rematch.py - Reads mapping against target sequences, checking mapping
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	7 and consensus sequences production
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	8 <https://github.com/B-UMMI/ReMatCh/>
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	9
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	10 Copyright (C) 2019 Miguel Machado <mpmachado@medicina.ulisboa.pt>
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	11
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	12 Last modified: August 08, 2019
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	13
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	14 This program is free software: you can redistribute it and/or modify
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	15 it under the terms of the GNU General Public License as published by
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	16 the Free Software Foundation, either version 3 of the License, or
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	17 (at your option) any later version.
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	18
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	19 This program is distributed in the hope that it will be useful,
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	20 but WITHOUT ANY WARRANTY; without even the implied warranty of
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	21 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	22 GNU General Public License for more details.
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	23
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	24 You should have received a copy of the GNU General Public License
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	25 along with this program. If not, see <http://www.gnu.org/licenses/>.
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	26 """
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	27
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	28 import os
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	29 import sys
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	30 import time
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	31 import argparse
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	32
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	33 try:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	34 from __init__ import __version__
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	35
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	36 import modules.utils as utils
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	37 import modules.seqFromWebTaxon as seq_from_web_taxon
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	38 import modules.download as download
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	39 import modules.rematch_module as rematch_module
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	40 import modules.checkMLST as check_mlst
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	41 except ImportError:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	42 from ReMatCh.__init__ import __version__
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	43
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	44 from ReMatCh.modules import utils as utils
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	45 from ReMatCh.modules import seqFromWebTaxon as seq_from_web_taxon
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	46 from ReMatCh.modules import download as download
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	47 from ReMatCh.modules import rematch_module as rematch_module
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	48 from ReMatCh.modules import checkMLST as check_mlst
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	49
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	50
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	51 def search_fastq_files(directory):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	52 files_extensions = ['.fastq.gz', '.fq.gz']
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	53 pair_end_files_separation = [['_R1_001.f', '_R2_001.f'], ['_1.f', '_2.f']]
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	54
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	55 list_ids = {}
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	56 directories = [d for d in os.listdir(directory) if
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	57 not d.startswith('.') and os.path.isdir(os.path.join(directory, d, ''))]
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	58 for directory_found in directories:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	59 if directory_found != 'pubmlst':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	60 directory_path = os.path.join(directory, directory_found, '')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	61
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	62 fastq_found = []
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	63 files = [f for f in os.listdir(directory_path) if
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	64 not f.startswith('.') and os.path.isfile(os.path.join(directory_path, f))]
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	65 for file_found in files:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	66 if file_found.endswith(tuple(files_extensions)):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	67 fastq_found.append(file_found)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	68
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	69 if len(fastq_found) == 1:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	70 list_ids[directory_found] = [os.path.join(directory_path, f) for f in fastq_found]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	71 elif len(fastq_found) >= 2:
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	72 file_pair = []
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	73
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	74 # Search pairs
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	75 for pe_separation in pair_end_files_separation:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	76 for fastq in fastq_found:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	77 if pe_separation[0] in fastq or pe_separation[1] in fastq:
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	78 file_pair.append(fastq)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	79
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	80 if len(file_pair) == 2:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	81 break
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	82 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	83 file_pair = []
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	84
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	85 # Search single
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	86 if len(file_pair) == 0:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	87 for pe_separation in pair_end_files_separation:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	88 for fastq in fastq_found:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	89 if pe_separation[0] not in fastq or pe_separation[1] not in fastq:
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	90 file_pair.append(fastq)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	91
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	92 if len(file_pair) >= 1:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	93 file_pair = file_pair[0]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	94
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	95 if len(file_pair) >= 1:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	96 list_ids[directory_found] = [os.path.join(directory_path, f) for f in file_pair]
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	97
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	98 return list_ids
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	99
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	100
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	101 def get_list_ids_from_file(file_list_ids):
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	102 list_ids = []
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	103
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	104 with open(file_list_ids, 'rtU') as lines:
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	105 for line in lines:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	106 line = line.rstrip('\r\n')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	107 if len(line) > 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	108 list_ids.append(line)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	109
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	110 if len(list_ids) == 0:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	111 sys.exit('No runIDs were found in ' + file_list_ids)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	112
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	113 return list_ids
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	114
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	115
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	116 def get_taxon_run_ids(taxon_name, outputfile):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	117 seq_from_web_taxon.run_seq_from_web_taxon(taxon_name, outputfile, True, True, True, False)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	118
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	119 run_ids = []
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	120 with open(outputfile, 'rtU') as reader:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	121 for line in reader:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	122 line = line.rstrip('\r\n')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	123 if len(line) > 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	124 if not line.startswith('#'):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	125 line = line.split('\t')
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	126 run_ids.append(line[0])
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	127
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	128 return run_ids
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	129
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	130
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	131 def get_list_ids(workdir, file_list_ids, taxon_name):
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	132 searched_fastq_files = False
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	133 list_ids = []
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	134 if file_list_ids is None and taxon_name is None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	135 list_ids = search_fastq_files(workdir)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	136 searched_fastq_files = True
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	137 elif file_list_ids is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	138 list_ids = get_list_ids_from_file(os.path.abspath(file_list_ids))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	139 elif taxon_name is not None and file_list_ids is None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	140 list_ids = get_taxon_run_ids(taxon_name, os.path.join(workdir, 'IDs_list.seqFromWebTaxon.tab'))
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	141
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	142 if len(list_ids) == 0:
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	143 sys.exit('No IDs were found')
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	144 return list_ids, searched_fastq_files
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	145
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	146
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	147 def format_gene_info(gene_specific_info, minimum_gene_coverage, minimum_gene_identity, reported_data_type, summary,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	148 sample, genes_present):
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	149 info = None
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	150 if gene_specific_info['gene_coverage'] >= minimum_gene_coverage and \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	151 gene_specific_info['gene_identity'] >= minimum_gene_identity:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	152 if summary and sample not in genes_present:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	153 genes_present[sample] = {}
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	154
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	155 if gene_specific_info['gene_number_positions_multiple_alleles'] == 0:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	156 s = str(gene_specific_info[reported_data_type])
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	157 info = str(s)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	158 if summary:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	159 genes_present[sample][gene_specific_info['header']] = str(s)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	160 else:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	161 s = 'multiAlleles_' + str(gene_specific_info[reported_data_type])
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	162 info = str(s)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	163 if summary:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	164 genes_present[sample][gene_specific_info['header']] = str(s)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	165 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	166 info = 'absent_' + str(gene_specific_info[reported_data_type])
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	167
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	168 return info, genes_present
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	169
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	170
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	171 def write_data_by_gene(gene_list_reference, minimum_gene_coverage, sample, data_by_gene, outdir, time_str, run_times,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	172 minimum_gene_identity, reported_data_type, summary, genes_present):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	173 combined_report = \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	174 os.path.join(outdir,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	175 'combined_report.data_by_gene.' + run_times + '.' + reported_data_type + '.' + time_str + '.tab')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	176
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	177 if reported_data_type == 'coverage_depth':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	178 reported_data_type = 'gene_mean_read_coverage'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	179 elif reported_data_type == 'sequence_coverage':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	180 reported_data_type = 'gene_coverage'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	181
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	182 combined_report_exist = os.path.isfile(combined_report)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	183 with open(combined_report, 'at') as writer:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	184 seq_list = sorted(gene_list_reference.keys())
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	185 if not combined_report_exist:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	186 writer.write('#sample' + '\t' + '\t'.join([gene_list_reference[seq] for seq in seq_list]) + '\n')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	187
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	188 results = {}
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	189 headers = []
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	190 for i in data_by_gene:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	191 results[data_by_gene[i]['header']], genes_present = format_gene_info(data_by_gene[i], minimum_gene_coverage,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	192 minimum_gene_identity,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	193 reported_data_type, summary, sample,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	194 genes_present)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	195 headers.append(data_by_gene[i]['header'])
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	196
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	197 if len(headers) != gene_list_reference:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	198 for gene in gene_list_reference:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	199 if gene not in headers:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	200 results[gene] = 'NA'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	201
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	202 writer.write(sample + '\t' + '\t'.join([results[seq] for seq in seq_list]) + '\n')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	203
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	204 return genes_present
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	205
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	206
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	207 def write_sample_report(sample, outdir, time_str, file_size, run_successfully_fastq, run_successfully_rematch_first,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	208 run_successfully_rematch_second, time_taken_fastq, time_taken_rematch_first,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	209 time_taken_rematch_second, time_taken_sample, sequencing_information, sample_data_general_first,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	210 sample_data_general_second, fastq_used):
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	211 sample_report = os.path.join(outdir, 'sample_report.' + time_str + '.tab')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	212 report_exist = os.path.isfile(sample_report)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	213
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	214 header_general = ['sample', 'sample_run_successfully', 'sample_run_time', 'files_size', 'download_run_successfully',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	215 'download_run_time', 'rematch_run_successfully_first', 'rematch_run_time_first',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	216 'rematch_run_successfully_second', 'rematch_run_time_second']
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	217 header_data_general = ['number_absent_genes', 'number_genes_multiple_alleles', 'mean_sample_coverage']
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	218 header_sequencing = ['run_accession', 'instrument_platform', 'instrument_model', 'library_layout', 'library_source',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	219 'extra_run_accession', 'nominal_length', 'read_count', 'base_count', 'date_download']
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	220
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	221 with open(sample_report, 'at') as writer:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	222 if not report_exist:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	223 writer.write('#' + '\t'.join(header_general) + '\t' + '_first\t'.join(header_data_general) + '_first\t' +
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	224 '_second\t'.join(header_data_general) + '_second\t' + '\t'.join(header_sequencing) + '\t' +
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	225 'fastq_used' + '\n')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	226
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	227 writer.write('\t'.join([sample,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	228 str(all([run_successfully_fastq is not False,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	229 run_successfully_rematch_first is not False,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	230 run_successfully_rematch_second is not False])),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	231 str(time_taken_sample),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	232 str(file_size),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	233 str(run_successfully_fastq),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	234 str(time_taken_fastq),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	235 str(run_successfully_rematch_first),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	236 str(time_taken_rematch_first),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	237 str(run_successfully_rematch_second),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	238 str(time_taken_rematch_second)]) +
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	239 '\t' + '\t'.join([str(sample_data_general_first[i]) for i in header_data_general]) +
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	240 '\t' + '\t'.join([str(sample_data_general_second[i]) for i in header_data_general]) +
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	241 '\t' + '\t'.join([str(sequencing_information[i]) for i in header_sequencing]) +
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	242 '\t' + ','.join(fastq_used) + '\n')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	243
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	244
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	245 def concatenate_extra_seq_2_consensus(consensus_sequence, reference_sequence, extra_seq_length, outdir):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	246 reference_dict, ignore, ignore = rematch_module.get_sequence_information(reference_sequence, extra_seq_length)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	247 consensus_dict, genes, ignore = rematch_module.get_sequence_information(consensus_sequence, 0)
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	248 number_consensus_with_sequences = 0
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	249 for k, values_consensus in list(consensus_dict.items()):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	250 for values_reference in list(reference_dict.values()):
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	251 if values_reference['header'] == values_consensus['header']:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	252 if len(set(consensus_dict[k]['sequence'])) > 1:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	253 number_consensus_with_sequences += 1
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	254 if extra_seq_length <= len(values_reference['sequence']):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	255 right_extra_seq = \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	256 '' if extra_seq_length == 0 else values_reference['sequence'][-extra_seq_length:]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	257 consensus_dict[k]['sequence'] = \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	258 values_reference['sequence'][:extra_seq_length] + \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	259 consensus_dict[k]['sequence'] + \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	260 right_extra_seq
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	261 consensus_dict[k]['length'] += extra_seq_length + len(right_extra_seq)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	262
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	263 consensus_concatenated = os.path.join(outdir, 'consensus_concatenated_extraSeq.fasta')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	264 with open(consensus_concatenated, 'wt') as writer:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	265 for i in consensus_dict:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	266 writer.write('>' + consensus_dict[i]['header'] + '\n')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	267 fasta_sequence_lines = rematch_module.chunkstring(consensus_dict[i]['sequence'], 80)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	268 for line in fasta_sequence_lines:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	269 writer.write(line + '\n')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	270
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	271 return consensus_concatenated, genes, consensus_dict, number_consensus_with_sequences
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	272
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	273
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	274 def clean_headers_reference_file(reference_file, outdir, extra_seq):
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	275 problematic_characters = ["\|", " ", ",", ".", "(", ")", "'", "/", ":"]
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	276 print('Checking if reference sequences contain ' + str(problematic_characters) + '\n')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	277 # headers_changed = False
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	278 new_reference_file = str(reference_file)
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	279 sequences, genes, headers_changed = rematch_module.get_sequence_information(reference_file, extra_seq)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	280 if headers_changed:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	281 print('At least one of the those characters was found. Replacing those with _' + '\n')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	282 new_reference_file = \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	283 os.path.join(outdir, os.path.splitext(os.path.basename(reference_file))[0] + '.headers_renamed.fasta')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	284 with open(new_reference_file, 'wt') as writer:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	285 for i in sequences:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	286 writer.write('>' + sequences[i]['header'] + '\n')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	287 fasta_sequence_lines = rematch_module.chunkstring(sequences[i]['sequence'], 80)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	288 for line in fasta_sequence_lines:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	289 writer.write(line + '\n')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	290 return new_reference_file, genes, sequences
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	291
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	292
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	293 def write_mlst_report(sample, run_times, consensus_type, st, alleles_profile, loci_order, outdir, time_str):
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	294 mlst_report = os.path.join(outdir, 'mlst_report.' + time_str + '.tab')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	295 mlst_report_exist = os.path.isfile(mlst_report)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	296 with open(mlst_report, 'at') as writer:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	297 if not mlst_report_exist:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	298 writer.write('\t'.join(['#sample', 'ReMatCh_run', 'consensus_type', 'ST'] + loci_order) + '\n')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	299 writer.write('\t'.join([sample, run_times, consensus_type, str(st)] + alleles_profile.split(',')) + '\n')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	300
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	301
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	302 def run_get_st(sample, mlst_dicts, consensus_sequences, mlst_consensus, run_times, outdir, time_str):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	303 if mlst_consensus == 'all':
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	304 for consensus_type in consensus_sequences:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	305 print('Searching MLST for ' + consensus_type + ' consensus')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	306 st, alleles_profile = check_mlst.get_st(mlst_dicts, consensus_sequences[consensus_type])
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	307 write_mlst_report(sample, run_times, consensus_type, st, alleles_profile, mlst_dicts[2], outdir, time_str)
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	308 print('ST found: ' + str(st) + ' (' + alleles_profile + ')')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	309 else:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	310 st, alleles_profile = check_mlst.get_st(mlst_dicts, consensus_sequences[mlst_consensus])
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	311 write_mlst_report(sample, run_times, mlst_consensus, st, alleles_profile, mlst_dicts[2], outdir, time_str)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	312 print('ST found for ' + mlst_consensus + ' consensus: ' + str(st) + ' (' + alleles_profile + ')')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	313
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	314
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	315 def write_summary_report(outdir, reported_data_type, time_str, gene_list_reference, genes_present):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	316 with open(os.path.join(outdir,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	317 'summary.{reported_data_type}.{time_str}.tab'.format(reported_data_type=reported_data_type,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	318 time_str=time_str)), 'wt') as writer:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	319 seq_list = []
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	320 for info in list(genes_present.values()):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	321 seq_list.extend(list(info.keys()))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	322 seq_list = list(set(seq_list))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	323 writer.write('#sample' + '\t' + '\t'.join([gene_list_reference[seq] for seq in sorted(seq_list)]) + '\n')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	324 for sample, info in list(genes_present.items()):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	325 data = []
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	326 for seq in sorted(seq_list):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	327 if seq in info:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	328 data.append(info[seq])
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	329 else:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	330 data.append('NF')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	331 writer.write(sample + '\t' + '\t'.join(data) + '\n')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	332
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	333
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	334 def run_rematch(args):
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	335 workdir = os.path.abspath(args.workdir)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	336 if not os.path.isdir(workdir):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	337 os.makedirs(workdir)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	338
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	339 aspera_key = os.path.abspath(args.asperaKey.name) if args.asperaKey is not None else None
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	340
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	341 # Start logger
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	342 logfile, time_str = utils.start_logger(workdir)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	343
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	344 # Get general information
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	345 script_path = utils.general_information(logfile, __version__, workdir, time_str, args.doNotUseProvidedSoftware,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	346 aspera_key, args.downloadCramBam, args.SRA, args.SRAopt)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	347
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	348 # Set list_ids
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	349 list_ids, searched_fastq_files = get_list_ids(workdir, args.listIDs.name if args.listIDs is not None else None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	350 args.taxon)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	351
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	352 mlst_sequences = None
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	353 mlst_dicts = None
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	354 if args.mlst is not None:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	355 time_taken_pub_mlst, mlst_dicts, mlst_sequences = check_mlst.download_pub_mlst_xml(args.mlst,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	356 args.mlstSchemaNumber,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	357 workdir)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	358 args.softClip_recodeRun = 'first'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	359
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	360 if args.reference is None:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	361 if args.mlst is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	362 reference_file = check_mlst.check_existing_schema(args.mlst, args.mlstSchemaNumber, script_path)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	363 args.extraSeq = 200
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	364 if reference_file is None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	365 print('It was not found provided MLST scheme sequences for ' + args.mlst)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	366 print('Trying to obtain reference MLST sequences from PubMLST')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	367 if len(mlst_sequences) > 0:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	368 reference_file = check_mlst.write_mlst_reference(args.mlst, mlst_sequences, workdir, time_str)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	369 args.extraSeq = 0
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	370 else:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	371 sys.exit('It was not possible to download MLST sequences from PubMLST!')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	372 else:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	373 print('Using provided scheme as referece: ' + reference_file)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	374 else:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	375 sys.exit('Need to provide at least one of the following options: "--reference" and "--mlst"')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	376 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	377 reference_file = os.path.abspath(args.reference.name)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	378
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	379 # Run ReMatCh for each sample
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	380 print('\n' + 'STARTING ReMatCh' + '\n')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	381
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	382 # Clean sequences headers
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	383 reference_file, gene_list_reference, reference_dict = clean_headers_reference_file(reference_file, workdir,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	384 args.extraSeq)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	385
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	386 if args.mlst is not None:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	387 problem_genes = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	388 for header in mlst_sequences:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	389 if header not in gene_list_reference:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	390 print('MLST gene {header} not found between reference sequences'.format(header=header))
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	391 problem_genes = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	392 if problem_genes:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	393 sys.exit('Missing MLST genes from reference sequences (at least sequences names do not match)!')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	394
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	395 if len(gene_list_reference) == 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	396 sys.exit('No sequences left')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	397
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	398 # To use in combined report
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	399
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	400 number_samples_successfully = 0
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	401 genes_present_coverage_depth = {}
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	402 genes_present_sequence_coverage = {}
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	403 for sample in list_ids:
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	404 sample_start_time = time.time()
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	405 print('\n\n' + 'Sample ID: ' + sample)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	406
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	407 # Create sample outdir
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	408 sample_outdir = os.path.join(workdir, sample, '')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	409 if not os.path.isdir(sample_outdir):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	410 os.mkdir(sample_outdir)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	411
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	412 run_successfully_fastq = None
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	413 time_taken_fastq = 0
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	414 sequencing_information = {'run_accession': None, 'instrument_platform': None, 'instrument_model': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	415 'library_layout': None, 'library_source': None, 'extra_run_accession': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	416 'nominal_length': None, 'read_count': None, 'base_count': None, 'date_download': None}
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	417 if not searched_fastq_files:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	418 # Download Files
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	419 time_taken_fastq, run_successfully_fastq, fastq_files, sequencing_information = \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	420 download.run_download(sample, args.downloadLibrariesType, aspera_key, sample_outdir,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	421 args.downloadCramBam, args.threads, args.downloadInstrumentPlatform, args.SRA,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	422 args.SRAopt)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	423 else:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	424 fastq_files = list_ids[sample]
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	425
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	426 file_size = None
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	427
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	428 run_successfully_rematch_first = None
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	429 run_successfully_rematch_second = None
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	430 time_taken_rematch_first = 0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	431 time_taken_rematch_second = 0
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	432 sample_data_general_first = None
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	433 sample_data_general_second = None
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	434 if run_successfully_fastq is not False:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	435 file_size = sum(os.path.getsize(fastq) for fastq in fastq_files)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	436 # Run ReMatCh
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	437 time_taken_rematch_first, run_successfully_rematch_first, data_by_gene, sample_data_general_first, \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	438 consensus_files, consensus_sequences = \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	439 rematch_module.run_rematch_module(sample, fastq_files, reference_file, args.threads, sample_outdir,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	440 args.extraSeq, args.minCovPresence, args.minCovCall,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	441 args.minFrequencyDominantAllele, args.minGeneCoverage,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	442 args.debug, args.numMapLoc, args.minGeneIdentity,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	443 'first', args.softClip_baseQuality, args.softClip_recodeRun,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	444 reference_dict, args.softClip_cigarFlagRecode,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	445 args.bowtieAlgo, args.bowtieOPT,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	446 gene_list_reference, args.notWriteConsensus, clean_run=True)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	447 if run_successfully_rematch_first:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	448 if args.mlst is not None and (args.mlstRun == 'first' or args.mlstRun == 'all'):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	449 run_get_st(sample, mlst_dicts, consensus_sequences, args.mlstConsensus, 'first', workdir, time_str)
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	450 genes_present_coverage_depth = write_data_by_gene(gene_list_reference, args.minGeneCoverage, sample,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	451 data_by_gene, workdir, time_str, 'first_run',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	452 args.minGeneIdentity, 'coverage_depth', args.summary,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	453 genes_present_coverage_depth)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	454 if args.reportSequenceCoverage:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	455 genes_present_sequence_coverage = write_data_by_gene(gene_list_reference, args.minGeneCoverage,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	456 sample, data_by_gene, workdir, time_str,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	457 'first_run', args.minGeneIdentity,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	458 'sequence_coverage', args.summary,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	459 genes_present_sequence_coverage)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	460 if args.doubleRun:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	461 rematch_second_outdir = os.path.join(sample_outdir, 'rematch_second_run', '')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	462 if not os.path.isdir(rematch_second_outdir):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	463 os.mkdir(rematch_second_outdir)
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	464 consensus_concatenated_fasta, consensus_concatenated_gene_list, consensus_concatenated_dict, \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	465 number_consensus_with_sequences = \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	466 concatenate_extra_seq_2_consensus(consensus_files['noMatter'], reference_file, args.extraSeq,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	467 rematch_second_outdir)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	468 if len(consensus_concatenated_gene_list) > 0:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	469 if args.mlst is None or \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	470 (args.mlst is not None and number_consensus_with_sequences == len(gene_list_reference)):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	471 time_taken_rematch_second, run_successfully_rematch_second, data_by_gene, \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	472 sample_data_general_second, consensus_files, consensus_sequences = \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	473 rematch_module.run_rematch_module(sample, fastq_files, consensus_concatenated_fasta,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	474 args.threads, rematch_second_outdir, args.extraSeq,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	475 args.minCovPresence, args.minCovCall,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	476 args.minFrequencyDominantAllele, args.minGeneCoverage,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	477 args.debug, args.numMapLoc,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	478 args.minGeneIdentity, 'second',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	479 args.softClip_baseQuality, args.softClip_recodeRun,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	480 consensus_concatenated_dict,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	481 args.softClip_cigarFlagRecode,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	482 args.bowtieAlgo, args.bowtieOPT,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	483 gene_list_reference, args.notWriteConsensus,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	484 clean_run=True)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	485 if not args.debug:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	486 os.remove(consensus_concatenated_fasta)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	487 if run_successfully_rematch_second:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	488 if args.mlst is not None and (args.mlstRun == 'second' or args.mlstRun == 'all'):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	489 run_get_st(sample, mlst_dicts, consensus_sequences, args.mlstConsensus, 'second',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	490 workdir, time_str)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	491 _ = write_data_by_gene(gene_list_reference, args.minGeneCoverage, sample, data_by_gene,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	492 workdir, time_str, 'second_run', args.minGeneIdentity,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	493 'coverage_depth', False, {})
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	494 if args.reportSequenceCoverage:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	495 _ = write_data_by_gene(gene_list_reference, args.minGeneCoverage, sample,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	496 data_by_gene, workdir, time_str, 'second_run',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	497 args.minGeneIdentity, 'sequence_coverage', False, {})
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	498 else:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	499 print('Some sequences missing after ReMatCh module first run. Second run will not be'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	500 ' performed')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	501 if os.path.isfile(consensus_concatenated_fasta):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	502 os.remove(consensus_concatenated_fasta)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	503 if os.path.isdir(rematch_second_outdir):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	504 utils.remove_directory(rematch_second_outdir)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	505 else:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	506 print('No sequences left after ReMatCh module first run. Second run will not be performed')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	507 if os.path.isfile(consensus_concatenated_fasta):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	508 os.remove(consensus_concatenated_fasta)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	509 if os.path.isdir(rematch_second_outdir):
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	510 utils.remove_directory(rematch_second_outdir)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	511
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	512 if not searched_fastq_files and not args.keepDownloadedFastq and fastq_files is not None:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	513 for fastq in fastq_files:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	514 if os.path.isfile(fastq):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	515 os.remove(fastq)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	516
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	517 time_taken = utils.run_time(sample_start_time)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	518
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	519 write_sample_report(sample, workdir, time_str, file_size, run_successfully_fastq,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	520 run_successfully_rematch_first, run_successfully_rematch_second, time_taken_fastq,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	521 time_taken_rematch_first, time_taken_rematch_second, time_taken, sequencing_information,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	522 sample_data_general_first if run_successfully_rematch_first else
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	523 {'number_absent_genes': None, 'number_genes_multiple_alleles': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	524 'mean_sample_coverage': None},
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	525 sample_data_general_second if run_successfully_rematch_second else
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	526 {'number_absent_genes': None, 'number_genes_multiple_alleles': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	527 'mean_sample_coverage': None},
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	528 fastq_files if fastq_files is not None else '')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	529
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	530 if all([run_successfully_fastq is not False,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	531 run_successfully_rematch_first is not False,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	532 run_successfully_rematch_second is not False]):
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	533 number_samples_successfully += 1
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	534
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	535 if args.summary:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	536 write_summary_report(workdir, 'coverage_depth', time_str, gene_list_reference, genes_present_coverage_depth)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	537 if args.reportSequenceCoverage:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	538 write_summary_report(workdir, 'sequence_coverage', time_str, gene_list_reference,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	539 genes_present_sequence_coverage)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	540
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	541 return number_samples_successfully, len(list_ids)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	542
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	543
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	544 def main():
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	545 if sys.version_info[0] < 3:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	546 sys.exit('Must be using Python 3. Try calling "python3 rematch.py"')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	547
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	548 parser = argparse.ArgumentParser(prog='rematch.py',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	549 description='Reads mapping against target sequences, checking mapping and'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	550 ' consensus sequences production',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	551 formatter_class=argparse.ArgumentDefaultsHelpFormatter)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	552 parser.add_argument('--version', help='Version information', action='version',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	553 version='{prog} v{version}'.format(prog=parser.prog, version=__version__))
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	554
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	555 parser_optional_general = parser.add_argument_group('General facultative options')
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	556 parser_optional_general.add_argument('-r', '--reference', type=argparse.FileType('r'),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	557 metavar='/path/to/reference_sequence.fasta',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	558 help='Fasta file containing reference sequences', required=False)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	559 parser_optional_general.add_argument('-w', '--workdir', type=str, metavar='/path/to/workdir/directory/',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	560 help='Path to the directory where ReMatCh will run and produce the outputs'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	561 ' with reads (ended with fastq.gz/fq.gz and, in case of PE data, pair-end'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	562 ' direction coded as _R1_001 / _R2_001 or _1 / _2) already'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	563 ' present (organized in sample folders) or to be downloaded',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	564 required=False, default='.')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	565 parser_optional_general.add_argument('-j', '--threads', type=int, metavar='N', help='Number of threads to use',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	566 required=False, default=1)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	567 parser_optional_general.add_argument('--mlst', type=str, metavar='"Streptococcus agalactiae"',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	568 help='Species name (same as in PubMLST) to be used in MLST'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	569 ' determination. ReMatCh will use Bowtie2 very-sensitive-local mapping'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	570 ' parameters and will recode the soft clip CIGAR flags of the first run',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	571 required=False)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	572 parser_optional_general.add_argument('--doNotUseProvidedSoftware', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	573 help='Tells ReMatCh to not use Bowtie2, Samtools and Bcftools that are'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	574 ' provided with it')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	575
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	576 parser_optional_download_exclusive = parser.add_mutually_exclusive_group()
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	577 parser_optional_download_exclusive.add_argument('-l', '--listIDs', type=argparse.FileType('r'),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	578 metavar='/path/to/list_IDs.txt',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	579 help='Path to list containing the IDs to be'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	580 ' downloaded (one per line)', required=False)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	581 parser_optional_download_exclusive.add_argument('-t', '--taxon', type=str, metavar='"Streptococcus agalactiae"',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	582 help='Taxon name for which ReMatCh will download fastq files',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	583 required=False)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	584
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	585 parser_optional_rematch = parser.add_argument_group('ReMatCh module facultative options')
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	586 parser_optional_rematch.add_argument('--extraSeq', type=int, metavar='N',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	587 help='Sequence length added to both ends of target sequences (usefull to'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	588 ' improve reads mapping to the target one) that will be trimmed in'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	589 ' ReMatCh outputs', required=False, default=0)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	590 parser_optional_rematch.add_argument('--minCovPresence', type=int, metavar='N',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	591 help='Reference position minimum coverage depth to consider the position to be'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	592 ' present in the sample', required=False, default=5)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	593 parser_optional_rematch.add_argument('--minCovCall', type=int, metavar='N',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	594 help='Reference position minimum coverage depth to perform a base call. Lower'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	595 ' coverage will be coded as N', required=False, default=10)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	596 parser_optional_rematch.add_argument('--minFrequencyDominantAllele', type=float, metavar='0.6',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	597 help='Minimum relative frequency of the dominant allele coverage depth (value'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	598 ' between [0, 1]). Positions with lower values will be considered as'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	599 ' having multiple alleles (and will be coded as N)', required=False,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	600 default=0.6)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	601 parser_optional_rematch.add_argument('--minGeneCoverage', type=int, metavar='N',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	602 help='Minimum percentage of target reference gene sequence covered'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	603 ' by --minCovPresence to consider a gene to be present (value'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	604 ' between [0, 100])', required=False, default=70)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	605 parser_optional_rematch.add_argument('--minGeneIdentity', type=int, metavar='N',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	606 help='Minimum percentage of identity of reference gene sequence covered'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	607 ' by --minCovCall to consider a gene to be present (value'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	608 ' between [0, 100]). One INDEL will be considered as one difference',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	609 required=False, default=80)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	610 parser_optional_rematch.add_argument('--numMapLoc', type=int, metavar='N', help=argparse.SUPPRESS, required=False,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	611 default=1)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	612 # parser_optional_rematch.add_argument('--numMapLoc', type=int, metavar='N', help='Maximum number of locations to which a read can map (sometimes useful when mapping against similar sequences)', required=False, default=1)
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	613 parser_optional_rematch.add_argument('--doubleRun', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	614 help='Tells ReMatCh to run a second time using as reference the noMatter'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	615 ' consensus sequence produced in the first run. This will improve'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	616 ' consensus sequence determination for sequences with high percentage of'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	617 ' target reference gene sequence covered')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	618 parser_optional_rematch.add_argument('--reportSequenceCoverage', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	619 help='Produce an extra combined_report.data_by_gene with the sequence coverage'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	620 ' instead of coverage depth')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	621 parser_optional_rematch.add_argument('--summary', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	622 help='Produce extra report files containing only sequences present in at least'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	623 ' one sample (usefull when using a large number of reference'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	624 ' sequences, and only for first run)')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	625 parser_optional_rematch.add_argument('--notWriteConsensus', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	626 help='Do not write consensus sequences')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	627 parser_optional_rematch.add_argument('--bowtieAlgo', type=str, metavar='"--very-sensitive-local"',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	628 help='Bowtie2 alignment mode. It can be an end-to-end alignment (unclipped'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	629 ' alignment) or local alignment (soft clipped alignment). Also, can'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	630 ' choose between fast or sensitive alignments. Please check Bowtie2'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	631 ' manual for extra'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	632 ' information: http://bowtie-bio.sourceforge.net/bowtie2/index.shtml .'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	633 ' This option should be provided between quotes and starting with'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	634 ' an empty space (like --bowtieAlgo " --very-fast") or using equal'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	635 ' sign (like --bowtieAlgo="--very-fast")',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	636 required=False, default='--very-sensitive-local')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	637 parser_optional_rematch.add_argument('--bowtieOPT', type=str, metavar='"--no-mixed"',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	638 help='Extra Bowtie2 options. This option should be provided between quotes and'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	639 ' starting with an empty space (like --bowtieOPT " --no-mixed") or using'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	640 ' equal sign (like --bowtieOPT="--no-mixed")',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	641 required=False)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	642 parser_optional_rematch.add_argument('--debug', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	643 help='DeBug Mode: do not remove temporary files')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	644
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	645 parser_optional_mlst = parser.add_argument_group('MLST facultative options')
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	646 parser_optional_rematch.add_argument('--mlstReference', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	647 help='If the curated scheme for MLST alleles is available, tells ReMatCh to'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	648 ' use these as reference (force Bowtie2 to run with very-sensitive-local'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	649 ' parameters, and sets --extraSeq to 200), otherwise ReMatCh uses the'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	650 ' first alleles of each MLST gene fragment in PubMLST as reference'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	651 ' sequences (force Bowtie2 to run with very-sensitive-local parameters,'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	652 ' and sets --extraSeq to 0)')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	653 parser_optional_mlst.add_argument('--mlstSchemaNumber', type=int, metavar='N',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	654 help='Number of the species PubMLST schema to be used in case of multiple schemes'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	655 ' available (by default will use the first schema)', required=False)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	656 parser_optional_mlst.add_argument('--mlstConsensus', choices=['noMatter', 'correct', 'alignment', 'all'], type=str,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	657 metavar='noMatter',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	658 help='Consensus sequence to be used in MLST'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	659 ' determination (available options: %(choices)s)', required=False,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	660 default='noMatter')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	661 parser_optional_mlst.add_argument('--mlstRun', choices=['first', 'second', 'all'], type=str, metavar='first',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	662 help='ReMatCh run outputs to be used in MLST determination (available'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	663 ' options: %(choices)s)', required=False, default='all')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	664
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	665 parser_optional_download = parser.add_argument_group('Download facultative options')
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	666 parser_optional_download.add_argument('-a', '--asperaKey', type=argparse.FileType('r'),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	667 metavar='/path/to/asperaweb_id_dsa.openssh',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	668 help='Tells ReMatCh to download fastq files from ENA using Aspera'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	669 ' Connect. With this option, the path to Private-key file'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	670 ' asperaweb_id_dsa.openssh must be provided (normaly found in'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	671 ' ~/.aspera/connect/etc/asperaweb_id_dsa.openssh).', required=False)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	672 parser_optional_download.add_argument('-k', '--keepDownloadedFastq', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	673 help='Tells ReMatCh to keep the fastq files downloaded')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	674 parser_optional_download.add_argument('--downloadLibrariesType', type=str, metavar='PAIRED',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	675 help='Tells ReMatCh to download files with specific library'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	676 ' layout (available options: %(choices)s)',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	677 choices=['PAIRED', 'SINGLE', 'BOTH'], required=False, default='BOTH')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	678 parser_optional_download.add_argument('--downloadInstrumentPlatform', type=str, metavar='ILLUMINA',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	679 help='Tells ReMatCh to download files with specific library layout (available'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	680 ' options: %(choices)s)', choices=['ILLUMINA', 'ALL'], required=False,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	681 default='ILLUMINA')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	682 parser_optional_download.add_argument('--downloadCramBam', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	683 help='Tells ReMatCh to also download cram/bam files and convert them to fastq'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	684 ' files')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	685
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	686 parser_optional_sra = parser.add_mutually_exclusive_group()
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	687 parser_optional_sra.add_argument('--SRA', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	688 help='Tells getSeqENA.py to download reads in fastq format only from NCBI SRA'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	689 ' database (not recommended)')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	690 parser_optional_sra.add_argument('--SRAopt', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	691 help='Tells getSeqENA.py to download reads from NCBI SRA if the download from ENA'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	692 ' fails')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	693
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	694 parser_optional_soft_clip = parser.add_argument_group('Soft clip facultative options')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	695 parser_optional_soft_clip.add_argument('--softClip_baseQuality', type=int, metavar='N',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	696 help='Base quality phred score in reads soft clipped regions',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	697 required=False,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	698 default=7)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	699 parser_optional_soft_clip.add_argument('--softClip_recodeRun', type=str, metavar='first',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	700 help='ReMatCh run to recode soft clipped regions (available'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	701 ' options: %(choices)s)', choices=['first', 'second', 'both', 'none'],
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	702 required=False, default='none')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	703 parser_optional_soft_clip.add_argument('--softClip_cigarFlagRecode', type=str, metavar='M',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	704 help='CIGAR flag to recode CIGAR soft clip (available options: %(choices)s)',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	705 choices=['M', 'I', 'X'], required=False, default='X')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	706
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	707 args = parser.parse_args()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	708
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	709 msg = []
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	710 if args.reference is None and not args.mlstReference:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	711 msg.append('At least --reference or --mlstReference should be provided')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	712 elif args.reference is not None and args.mlstReference:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	713 msg.append('Only --reference or --mlstReference should be provided')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	714 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	715 if args.mlstReference:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	716 if args.mlst is None:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	717 msg.append('Please provide species name using --mlst')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	718 if args.minFrequencyDominantAllele < 0 or args.minFrequencyDominantAllele > 1:
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	719 msg.append('--minFrequencyDominantAllele should be a value between [0, 1]')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	720 if args.minGeneCoverage < 0 or args.minGeneCoverage > 100:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	721 msg.append('--minGeneCoverage should be a value between [0, 100]')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	722 if args.minGeneIdentity < 0 or args.minGeneIdentity > 100:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	723 msg.append('--minGeneIdentity should be a value between [0, 100]')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	724 if args.notWriteConsensus and args.doubleRun:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	725 msg.append('--notWriteConsensus and --doubleRun cannot be used together.'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	726 ' Maybe you only want to use --doubleRun')
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	727
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	728 if len(msg) > 0:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	729 argparse.ArgumentParser.error('\n'.join(msg))
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	730
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	731 start_time = time.time()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	732
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	733 number_samples_successfully, samples_total_number = run_rematch(args)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	734
3 0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	735 print('\n' + 'END ReMatCh')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	736 print('\n' +
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	737 str(number_samples_successfully) + ' samples out of ' + str(samples_total_number) + ' run successfully')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: 0 diff changeset	738 time_taken = utils.run_time(start_time)
0 965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	739 del time_taken
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	740
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	741 if number_samples_successfully == 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	742 sys.exit('No samples run successfully!')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	743
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	744
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	745 if __name__ == "__main__":
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8 cstrittmatter parents: diff changeset	746 main()

Mercurial > repos > cstrittmatter > test_eurl_vtec_wgs_pt

annotate scripts/ReMatCh/rematch.py @ 3:0cbed1c0a762 draft default tip