Mercurial > repos > cstrittmatter > test_eurl_vtec_wgs_pt

annotate scripts/modules/typing.py @ 3:0cbed1c0a762 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
author cstrittmatter
date Tue, 28 Jan 2020 10:42:31 -0500
parents 965517909457
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
1 import os.path
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
2 import functools
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
3
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
4 try:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
5 import modules.utils as utils
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
6 except ImportError:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
7 from pathotyping.modules import utils as utils
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
8
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
9
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
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}}
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
11
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
12
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
13 def simplify_data_by_gene(data_by_gene):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
14 cleaned_data_by_gene = {}
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
15 for counter, data in list(data_by_gene.items()):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
16 cleaned_data_by_gene[data['header'].lower()] = {'gene_identity': data['gene_identity'],
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
17 'gene_coverage': data['gene_coverage'],
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
18 'gene_depth': data['gene_mean_read_coverage']}
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
19 return cleaned_data_by_gene
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
20
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
21
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
22 def possible_types(data_by_gene, typing_rules_file, min_gene_coverage, min_gene_identity, min_gene_depth):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
23 data_by_gene = simplify_data_by_gene(data_by_gene)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
24
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
25 possible_pathotypes = []
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
26 genes_present = []
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
27 with open(typing_rules_file, 'rtU') as reader:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
28 genes = []
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
29 for line in reader:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
30 line = line.splitlines()[0]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
31 if len(line) > 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
32 line = line.split('\t')
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
33 if not line[0].startswith('##'):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
34 if line[0].startswith('#'):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
35 genes = line[1:]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
36 else:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
37 profile = line[1:]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
38 congruence = []
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
39 for x, gene_requirement in enumerate(profile):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
40 if data_by_gene[genes[x].lower()]['gene_coverage'] >= min_gene_coverage and \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
41 data_by_gene[genes[x].lower()]['gene_identity'] >= min_gene_identity and \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
42 data_by_gene[genes[x].lower()]['gene_depth'] >= min_gene_depth:
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
43 gene_present = True
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
44 genes_present.append(genes[x])
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
45 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
46 gene_present = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
47
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
48 gene_requirement = \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
49 True if gene_requirement == '1' else False if gene_requirement == '0' else None
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
50 if gene_requirement is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
51 if gene_present == gene_requirement:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
52 congruence.append(True)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
53 else:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
54 congruence.append(False)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
55 else:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
56 congruence.append(True)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
57 if all(congruence):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
58 possible_pathotypes.append(line[0])
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
59 return possible_pathotypes, list(set(genes_present))
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
60
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
61
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
62 module_timer = functools.partial(utils.timer, name='Module Typing')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
63
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
64
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
65 @module_timer
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
66 def typing(data_by_gene, typing_rules_file, min_gene_coverage, min_gene_identity, min_gene_depth, outdir):
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
67 possible_pathotypes, genes_present = possible_types(data_by_gene, typing_rules_file, min_gene_coverage,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
68 min_gene_identity, min_gene_depth)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
69 with open(os.path.join(outdir, 'patho_typing.report.txt'), 'wt') as writer_report:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
70 with open(os.path.join(outdir, 'patho_typing.extended_report.txt'), 'wt') as writer_extended_report:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
71 writer_extended_report.write('#Pathotypes_found' + '\n')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
72 if len(possible_pathotypes) > 0:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
73 writer_report.write('\n'.join(possible_pathotypes) + '\n')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
74 writer_extended_report.write('\n'.join(possible_pathotypes) + '\n')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
75 print('\n' + 'Pathotypes found:' + '\n')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
76 print('\n'.join(possible_pathotypes) + '\n')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
77 else:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
78 writer_report.write('TND' + '\n') # Type Not Determined
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
79 writer_extended_report.write('TND' + '\n') # Type Not Determined
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
80 print('\n' + 'It was not possible to identify any possible pathotype match' + '\n')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
81 writer_extended_report.write('\n' + '#Genes_present' + '\n')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
82 writer_extended_report.write('\n'.join(genes_present) + '\n')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
83
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
84 return None, None