Mercurial > repos > cstrittmatter > test_eurl_vtec_wgs_pt
annotate EURL_VTEC_WGS_PT.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 # -*- coding: utf-8 -*- |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
3 """ |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
4 ############################################################################ |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
5 # Istituto Superiore di Sanita' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
6 # European Union Reference Laboratory (EU-RL) for Escherichia coli, including Verotoxigenic E. coli (VTEC) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
7 # Developer: Arnold Knijn arnold.knijn@iss.it |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
8 ############################################################################ |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
9 """ |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
10 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
11 import argparse |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
12 import sys |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
13 import os |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
14 import HTML |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
15 import datetime |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
16 import fileinput |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
17 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
18 BASE_URL = 'https://galaxytrakr.org' |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
19 TOOL_DIR = os.path.dirname(os.path.abspath(__file__)) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
20 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
21 def insertFile(filename, report): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
22 with open(filename) as html_in: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
23 for line in html_in: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
24 report.write(line) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
25 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
26 def insertFileAsTable(filename, report, hasheader=False, tabclass="table table-rep"): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
27 with open(filename) as table_in: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
28 table_data = [[str(col) for col in row.split('\t')] for row in table_in] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
29 insertTable(table_data, report, hasheader, tabclass) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
30 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
31 def insertTable(table_data, report, hasheader=False, tabclass="table table-rep"): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
32 if hasheader: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
33 htmlcode = HTML.table(table_data[1:], attribs={'class':tabclass}, header_row=table_data[0]) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
34 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
35 htmlcode = HTML.table(table_data, attribs={'class':tabclass}) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
36 report.write(htmlcode) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
37 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
38 def openFileAsTable(filename): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
39 with open(filename) as table_in: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
40 table_data = [[str(col) for col in row.split('\t')] for row in table_in] |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
41 return table_data |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
42 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
43 def __main__(): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
44 parser = argparse.ArgumentParser() |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
45 parser.add_argument('--serotyping', dest='serotyping', help='perform serotyping', action='store_true') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
46 parser.add_argument('--virulotyping', dest='virulotyping', help='perform virulotyping', action='store_true') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
47 parser.add_argument('--shigatoxintyping', dest='shigatoxintyping', help='perform shigatoxintyping', action='store_true') |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
48 parser.add_argument('--amrtyping', dest='amrtyping', help='perform amrtyping', action='store_true') |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
49 parser.add_argument('-1', '--input1', dest='input1', help='forward or single-end reads file in Sanger FASTQ format') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
50 parser.add_argument('--input1_ext', dest='input1_ext', help='extension of forward or single-end reads file in Sanger FASTQ format') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
51 parser.add_argument('--input1_name', dest='input1_name', help='name of forward or single-end reads file in Sanger FASTQ format') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
52 parser.add_argument('-2', '--input2', dest='input2', help='reverse reads file in Sanger FASTQ format') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
53 parser.add_argument('--input2_ext', dest='input2_ext', help='extension of reverse reads file in Sanger FASTQ format') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
54 parser.add_argument('--input2_name', dest='input2_name', help='name of reverse reads file in Sanger FASTQ format') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
55 parser.add_argument('--html1', dest='html1', help='html FASTQC file') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
56 parser.add_argument('--html1_id', dest='html1_id', help='html FASTQC file id') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
57 parser.add_argument('--html1_path', dest='html1_path', help='html FASTQC file path') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
58 parser.add_argument('--text1', dest='text1', help='text FASTQC file') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
59 parser.add_argument('--html2', dest='html2', help='html FASTQC file') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
60 parser.add_argument('--html2_id', dest='html2_id', help='html FASTQC file id') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
61 parser.add_argument('--html2_path', dest='html2_path', help='html FASTQC file path') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
62 parser.add_argument('--text2', dest='text2', help='text FASTQC file') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
63 parser.add_argument('--log', dest='logfile', help='log file') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
64 parser.add_argument('--virulotyper', dest='virulotyper', help='Virulotyping Mapping reads') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
65 parser.add_argument('--virulotyper_id', dest='virulotyper_id', help='Virulotyping Mapping reads id') |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
66 parser.add_argument('--stx', dest='stx', help='Shiga toxin') |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
67 parser.add_argument('--mlstsevenloci', dest='mlstsevenloci', help='Multi Locus Alleles table') |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
68 parser.add_argument('--amr', dest='amr', help='SPAdes log') |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
69 parser.add_argument('--amr_id', dest='amr_id', help='AMR file id') |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
70 parser.add_argument('--antigen_O', dest='antigen_O', help='Antigen for O') |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
71 parser.add_argument('--antigen_H', dest='antigen_H', help='Antigen for H') |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
72 parser.add_argument('--output', dest='output', help='output report html file') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
73 args = parser.parse_args() |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
74 |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
75 log = open(args.logfile, 'w') |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
76 log.write("EURL VTEC WGS PT v3.0\n\nTool versions\n=============\n") |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
77 if args.input2: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
78 # FASTQC |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
79 os.system("python " + TOOL_DIR + "/scripts/rgFastQC.py -i " + args.input1 + " -d " + args.html1_path + " -o " + args.html1 + " -t " + args.text1 + " -f " + args.input1_ext + " -j " + args.input1_name + " -e " + "fastqc") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
80 os.system("rm -r " + args.html1_path) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
81 os.system("python " + TOOL_DIR + "/scripts/rgFastQC.py -i " + args.input2 + " -d " + args.html2_path + " -o " + args.html2 + " -t " + args.text2 + " -f " + args.input2_ext + " -j " + args.input2_name + " -e " + "fastqc") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
82 os.system("rm -r " + args.html2_path) |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
83 log.write(os.popen("fastqc -v").read()) |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
84 # TRIMMING |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
85 os.system("python " + TOOL_DIR + "/scripts/fastq_positional_quality_trimming.py -1 " + args.input1 + " --maxlt 300 --lt 17 --rt 0 --minqt 25 --avgqt 27.0 --minlf -1 --trimmed1 input_t1.fq --log trimming_logfile -2 " + args.input2 + " --trimmed2 input_t2.fq --trimmedunpaired trimmedunpaired") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
86 log.write("\nfastq_positional_quality_trimming v1.0\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
87 log.write("parameters: maxlt=300, lt=17, rt=0, minqt=25, avgqt=27.0, minlf=-1\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
88 if args.shigatoxintyping or args.serotyping: |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
89 # ASSEMBLY |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
90 os.system("perl " + TOOL_DIR + "/scripts/spades.pl contigs.fa spades_contig_stats spades_scaffolds spades_scaffold_stats spades_log NODE spades.py --disable-gzip-output --isolate -t \${GALAXY_SLOTS:-16} --pe1-ff --pe1-1 fastq:input_t1.fq --pe1-2 fastq:input_t2.fq") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
91 log.write(os.popen("spades.py -v").read()) |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
92 log.write("parameters: --isolate, pe1-ff, pe1-1, pe1-2\n\n") |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
93 if args.virulotyping: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
94 # VIRULOTYPER |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
95 os.system("perl " + TOOL_DIR + "/scripts/patho_typing.pl 'python " + TOOL_DIR + "/scripts/patho_typing.py -s Escherichia coli -f " + args.input1 + " " + args.input2 + " -o output_dir -j 1 --minGeneCoverage 90 --minGeneIdentity 90 --minGeneDepth 15'") |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
96 os.system("cat pathotyper_rep_tot_tab > " + args.virulotyper) |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
97 log.write("\n\nViruloTyper\n===========\npatho_typing v1.0\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
98 log.write("parameters: minGeneCoverage=90, minGeneIdentity=90, minGeneDepth=15\n\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
99 log.write(os.popen("cat " + TOOL_DIR + "/data/ViruloTyping_db.txt").read()) |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
100 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
101 # FASTQC |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
102 os.system("python " + TOOL_DIR + "/scripts/rgFastQC.py -i " + args.input1 + " -d " + args.html1_path + " -o " + args.html1 + " -t " + args.text1 + " -f " + args.input1_ext + " -j " + args.input1_name + " -e " + "fastqc") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
103 os.system("rm -r " + args.html1_path) |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
104 log.write(os.popen("fastqc -v").read()) |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
105 # TRIMMING |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
106 os.system("python " + TOOL_DIR + "/scripts/fastq_positional_quality_trimming.py -1 " + args.input1 + " --maxlt 360 --lt 10 --rt 0 --minqt 25 --avgqt 27.0 --minlf 50 --trimmed1 input_t1.fq --log trimming_logfile") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
107 log.write("\nfastq_positional_quality_trimming v1.0\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
108 log.write("parameters: maxlt=360, lt=10, rt=0, minqt=25, avgqt=27.0, minlf=50\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
109 if args.shigatoxintyping or args.serotyping: |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
110 # ASSEMBLY |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
111 os.system("perl " + TOOL_DIR + "/scripts/spades.pl contigs.fa spades_contig_stats spades_scaffolds spades_scaffold_stats spades_log NODE spades.py --disable-gzip-output --isolate -t ${GALAXY_SLOTS:-16} --iontorrent -s fastq:input_t1.fq") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
112 log.write(os.popen("spades.py -v").read()) |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
113 log.write("parameters: --isolate, --iontorrent\n\n") |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
114 if args.virulotyping: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
115 # VIRULOTYPER |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
116 os.system("perl " + TOOL_DIR + "/scripts/patho_typing.pl 'python " + TOOL_DIR + "/scripts/patho_typing.py -s Escherichia coli -f " + args.input1 + " -o output_dir -j 1 --minGeneCoverage 90 --minGeneIdentity 90 --minGeneDepth 15'") |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
117 os.system("cat pathotyper_rep_tot_tab > " + args.virulotyper) |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
118 log.write("\n\nViruloTyper\n===========\npatho_typing v1.0\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
119 log.write("parameters: minGeneCoverage=90, minGeneIdentity=90, minGeneDepth=15\n\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
120 log.write(os.popen("cat " + TOOL_DIR + "/data/ViruloTyping_db.txt").read()) |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
121 # SEQUENCETYPER |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
122 if args.input2: |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
123 os.system("mentalist call --output_votes -o 'mentalist_out' --db '" + TOOL_DIR + "/data/escherichia_coli_pubmlst_k31_2018-10-09/escherichia_coli_pubmlst_k31_m023_2018-10-09.jld' -1 input_t1.fq -2 input_t2.fq") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
124 else: |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
125 os.system("mentalist call --output_votes -o 'mentalist_out' --db '" + TOOL_DIR + "/data/escherichia_coli_pubmlst_k31_2018-10-09/escherichia_coli_pubmlst_k31_m023_2018-10-09.jld' -1 input_t1.fq") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
126 os.system("mv mentalist_out.byvote " + args.mlstsevenloci) |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
127 sequence_typing = openFileAsTable(args.mlstsevenloci) |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
128 sequence_qc = openFileAsTable("mentalist_out.coverage.txt") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
129 log.write("\n\nSequenceTyper\n=============\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
130 log.write(os.popen("mentalist -v | grep MentaLiST").read()) |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
131 log.write("\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
132 log.write(os.popen("cat " + TOOL_DIR + "/data/SequenceTyping_db.txt").read()) |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
133 if args.shigatoxintyping: |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
134 # SHIGATOXIN TYPER |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
135 os.system("echo 'sseqid\tpident\tlength\tpositive' > mmseqs_shigatoxin_fct") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
136 os.system("mmseqs easy-search --search-type 3 --format-output query,target,pident,alnlen,tlen contigs.fa " + TOOL_DIR + "/data/stxDB mmseqs_STX /tmp") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
137 os.system(TOOL_DIR + "/scripts/extract_shigatoxin.sh") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
138 os.system("cat mmseqs_shigatoxin_fc >> mmseqs_shigatoxin_fct") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
139 shigatoxin_typing = openFileAsTable("mmseqs_shigatoxin_fc") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
140 log.write("\n\nShigatoxinTyper\n=============\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
141 log.write(os.popen("cat " + TOOL_DIR + "/data/ShigatoxinTyping_db.txt").read()) |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
142 if args.serotyping: |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
143 # SEROTYPER |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
144 os.system("echo 'sseqid\tpident\tlength\tpositive' > mmseqs_OH_fc") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
145 # SEROTYPER O |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
146 os.system("mmseqs easy-search --search-type 3 --format-output target,pident,alnlen,tlen contigs.fa " + TOOL_DIR + "/data/O_typeDB mmseqs_O /tmp") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
147 os.system("awk -F '\t' '($3>800 && $4>800) { print $1 FS $2 FS $3 FS $4 }' mmseqs_O | sort -nrk 2 -nrk 3 > mmseqs_O_fc") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
148 sero_typing_o = openFileAsTable("mmseqs_O_fc") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
149 os.system("cat mmseqs_O_fc >> mmseqs_OH_fc") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
150 # SEROTYPER H |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
151 os.system("mmseqs easy-search --search-type 3 --format-output target,pident,alnlen,tlen contigs.fa " + TOOL_DIR + "/data/H_typeDB mmseqs_H /tmp") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
152 os.system("awk -F '\t' '($3>800 && $4>800) { print $1 FS $2 FS $3 FS $4 }' mmseqs_H | sort -nrk 2 -nrk 3 > mmseqs_H_fc") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
153 sero_typing_h = openFileAsTable("mmseqs_H_fc") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
154 os.system("cat mmseqs_H_fc >> mmseqs_OH_fc") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
155 if os.stat('mmseqs_O_fc').st_size == 0 and os.stat('mmseqs_H_fc').st_size == 0: |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
156 os.system("echo '-\t-\t-\t-' >> mmseqs_OH_fc") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
157 log.write("\n\nSeroTyper\n=============\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
158 log.write(os.popen("cat " + TOOL_DIR + "/data/SeroTyping_db.txt").read()) |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
159 if args.amrtyping: |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
160 # AMRGENES |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
161 os.system("amrfinder --threads 4 --database " + TOOL_DIR + "/data/amrfinder -n contigs.fa -O Escherichia -o " + args.amr) |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
162 # REPORT |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
163 try: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
164 report = open(args.output, 'w') |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
165 # write head html |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
166 insertFile(TOOL_DIR + "/report_head.html", report) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
167 report.write("<td><h1>EURL VTEC WGS PT</h1><h2>Report for %s</h2>%s</td>" % (args.input1_name, datetime.datetime.utcnow().strftime("%Y-%m-%d %H:%M UTC"))) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
168 insertFile(TOOL_DIR + "/report_head2.html", report) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
169 # write results |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
170 report.write("<h3>Summary</h3>\n") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
171 if args.serotyping: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
172 report.write("<p>Serotype: ") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
173 if len(sero_typing_o) == 0: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
174 report.write("O?") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
175 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
176 report.write("%s" % sero_typing_o[0][0][sero_typing_o[0][0].rfind("O"):]) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
177 if len(sero_typing_h) == 0: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
178 report.write(":H?") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
179 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
180 report.write(":%s" % sero_typing_h[0][0][sero_typing_h[0][0].rfind("H"):]) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
181 report.write("</p>\n") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
182 report.write("<p>Sequence type: ") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
183 if len(sequence_typing) < 2: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
184 report.write("Sequence typing failed") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
185 elif sequence_typing[1][1] == "failed": |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
186 report.write("Sequence typing failed") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
187 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
188 report.write("ST%s" % sequence_typing[1][8]) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
189 report.write("</p>\n") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
190 if args.virulotyping: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
191 os.system("sort " + args.virulotyper + " | awk '/eae_|stx1._|stx2._|ehxa_/ && $2>50 && !seen[substr($1, 1, index($1, \"_\")-2)]++ { printf(\"%s%s\",sep,substr($1, 1, index($1, \"_\")-1));sep=\", \" }END{print \"\"}' > virulotyper_rep") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
192 for line in fileinput.input("virulotyper_rep", inplace=True): |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
193 print(line.replace("1a", "1"),) |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
194 for line in fileinput.input("virulotyper_rep", inplace=True): |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
195 print(line.replace("2a", "2"),) |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
196 for line in fileinput.input("virulotyper_rep", inplace=True): |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
197 print(line.replace("1b", "1"),) |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
198 for line in fileinput.input("virulotyper_rep", inplace=True): |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
199 print(line.replace("2b", "2"),) |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
200 report.write("<p>Virulotypes: ") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
201 insertFile("virulotyper_rep", report) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
202 report.write("</p>\n") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
203 if args.shigatoxintyping: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
204 report.write("<p>Stx Subtypes: ") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
205 if len(shigatoxin_typing) == 0: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
206 report.write("No subtype match found") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
207 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
208 shigatoxin_subtype = "" |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
209 shigatoxin_types = openFileAsTable(TOOL_DIR + "/data/stx_subtypes") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
210 for subtype in shigatoxin_typing: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
211 blast_pident_100 = float(subtype[1]) == 100 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
212 if (blast_pident_100): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
213 for item in shigatoxin_types: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
214 if item[0] == subtype[0]: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
215 shigatoxin_subtype = item[1] + " " + shigatoxin_subtype |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
216 if len(shigatoxin_subtype) == 0: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
217 shigatoxin_subtype = "No complete subtype match found" |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
218 report.write("%s" % shigatoxin_subtype) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
219 report.write("</p>\n") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
220 # Quality Check |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
221 disclaimer = False |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
222 sequence_qc_cov = 0 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
223 for x in range(7): |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
224 if float(sequence_qc[x+1][2]) < 1: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
225 disclaimer = True |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
226 sequence_qc_cov = sequence_qc_cov + float(sequence_qc[x+1][3]) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
227 if sequence_qc_cov < 210: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
228 disclaimer = True |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
229 if disclaimer: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
230 report.write("<p style='font-weight:bold;color:red'>Disclaimer: The data analysed do not fulfill minimum quality parameters, please consider repeating the sequencing.</p>\n") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
231 report.write("<hr/><h3>Raw data quality check</h3>\n") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
232 if args.input2: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
233 report.write("<p>FASTQC result forward: <a href='%s/datasets/%s/display/?preview=True'>Webpage</a></p>\n" % (BASE_URL, args.html1_id)) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
234 report.write("<p>FASTQC result reverse: <a href='%s/datasets/%s/display/?preview=True'>Webpage</a></p>\n" % (BASE_URL, args.html2_id)) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
235 else: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
236 report.write("<p>FASTQC result: <a href='%s/datasets/%s/display/?preview=True'>Webpage</a></p>\n" % (BASE_URL, args.html1_id)) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
237 if args.serotyping: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
238 report.write("<br/><hr/><h3>Serotyping</h3>\n") |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
239 insertFileAsTable("mmseqs_OH_fc", report, True) |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
240 report.write("<br/><hr/><h3>Multi Locus Sequence Typing</h3>\n") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
241 if len(sequence_typing) > 1: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
242 insertTable(sequence_typing, report, True) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
243 if args.virulotyping: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
244 report.write("<br/><hr/><h3>Virulotyping</h3>\n") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
245 report.write("<p>This table is filtered for results with >90%% gene coverage, unfiltered results can be found <a href='%s/datasets/%s/display/?preview=True'>here</a></p>\n" % (BASE_URL, args.virulotyper_id)) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
246 insertFileAsTable("pathotyper_rep_tab", report, True, "table table-cross") |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
247 if args.shigatoxintyping: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
248 report.write("<br/><hr/><h3>Shiga toxin typing</h3>\n") |
3
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
249 insertFileAsTable("mmseqs_shigatoxin_fct", report, True) |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
250 if args.amrtyping: |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
251 report.write("<br/><hr/><h3>AMR typing</h3>\n") |
0cbed1c0a762
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
0
diff
changeset
|
252 report.write("<p>AMR result: <a href='%s/datasets/%s/display/?preview=True'>Webpage</a></p>\n" % (BASE_URL, args.amr_id)) |
0
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
253 # write tail html |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
254 insertFile(TOOL_DIR + "/report_tail.html", report) |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
255 finally: |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
256 report.close() |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
257 |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
258 if __name__ == "__main__": |
965517909457
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff
changeset
|
259 __main__() |