Mercurial > repos > iss > eurl_vtec_wgs_pt
annotate scripts/modules/typing.py @ 6:20ff3dca457f draft default tip
planemo upload commit 6857c749c21f580c828aba3543e294b69d32b662
author | iss |
---|---|
date | Mon, 23 Oct 2023 11:45:36 +0000 (18 months ago) |
parents | c6bab5103a14 |
children |
rev | line source |
---|---|
0
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
1 import os.path |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
2 import functools |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
3 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
4 try: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
5 import modules.utils as utils |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
6 except ImportError: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
7 from pathotyping.modules import utils as utils |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
8 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
9 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
10 # {1: {'gene_identity': 0, 'gene_mean_read_coverage': 0.0, 'gene_number_positions_multiple_alleles': 0, 'header': 'fyuA', 'gene_coverage': 0.0, 'gene_low_coverage': 100.0}} |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
11 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
12 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
13 def simplify_data_by_gene(data_by_gene): |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
14 cleaned_data_by_gene = {} |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
15 for counter, data in list(data_by_gene.items()): |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
16 cleaned_data_by_gene[data['header'].lower()] = {'gene_identity': data['gene_identity'], |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
17 'gene_coverage': data['gene_coverage'], |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
18 'gene_depth': data['gene_mean_read_coverage']} |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
19 return cleaned_data_by_gene |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
20 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
21 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
22 def possible_types(data_by_gene, typing_rules_file, min_gene_coverage, min_gene_identity, min_gene_depth): |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
23 data_by_gene = simplify_data_by_gene(data_by_gene) |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
24 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
25 possible_pathotypes = [] |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
26 genes_present = [] |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
27 with open(typing_rules_file, 'rtU') as reader: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
28 genes = [] |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
29 for line in reader: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
30 line = line.splitlines()[0] |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
31 if len(line) > 0: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
32 line = line.split('\t') |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
33 if not line[0].startswith('##'): |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
34 if line[0].startswith('#'): |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
35 genes = line[1:] |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
36 else: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
37 profile = line[1:] |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
38 congruence = [] |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
39 for x, gene_requirement in enumerate(profile): |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
40 if data_by_gene[genes[x].lower()]['gene_coverage'] >= min_gene_coverage and \ |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
41 data_by_gene[genes[x].lower()]['gene_identity'] >= min_gene_identity and \ |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
42 data_by_gene[genes[x].lower()]['gene_depth'] >= min_gene_depth: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
43 gene_present = True |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
44 genes_present.append(genes[x]) |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
45 else: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
46 gene_present = False |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
47 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
48 gene_requirement = \ |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
49 True if gene_requirement == '1' else False if gene_requirement == '0' else None |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
50 if gene_requirement is not None: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
51 if gene_present == gene_requirement: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
52 congruence.append(True) |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
53 else: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
54 congruence.append(False) |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
55 else: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
56 congruence.append(True) |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
57 if all(congruence): |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
58 possible_pathotypes.append(line[0]) |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
59 return possible_pathotypes, list(set(genes_present)) |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
60 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
61 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
62 module_timer = functools.partial(utils.timer, name='Module Typing') |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
63 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
64 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
65 @module_timer |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
66 def typing(data_by_gene, typing_rules_file, min_gene_coverage, min_gene_identity, min_gene_depth, outdir): |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
67 possible_pathotypes, genes_present = possible_types(data_by_gene, typing_rules_file, min_gene_coverage, |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
68 min_gene_identity, min_gene_depth) |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
69 with open(os.path.join(outdir, 'patho_typing.report.txt'), 'wt') as writer_report: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
70 with open(os.path.join(outdir, 'patho_typing.extended_report.txt'), 'wt') as writer_extended_report: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
71 writer_extended_report.write('#Pathotypes_found' + '\n') |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
72 if len(possible_pathotypes) > 0: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
73 writer_report.write('\n'.join(possible_pathotypes) + '\n') |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
74 writer_extended_report.write('\n'.join(possible_pathotypes) + '\n') |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
75 print('\n' + 'Pathotypes found:' + '\n') |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
76 print('\n'.join(possible_pathotypes) + '\n') |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
77 else: |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
78 writer_report.write('TND' + '\n') # Type Not Determined |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
79 writer_extended_report.write('TND' + '\n') # Type Not Determined |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
80 print('\n' + 'It was not possible to identify any possible pathotype match' + '\n') |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
81 writer_extended_report.write('\n' + '#Genes_present' + '\n') |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
82 writer_extended_report.write('\n'.join(genes_present) + '\n') |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
83 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
84 return None, None |