annotate correct.py @ 18:e4d75f9efb90 draft

planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
author nick
date Thu, 02 Feb 2017 18:44:31 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
18
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
1 #!/usr/bin/env python
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
2 from __future__ import division
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
3 from __future__ import print_function
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
4 import os
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
5 import sys
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
6 import gzip
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
7 import logging
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
8 import argparse
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
9 import resource
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
10 import subprocess
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
11 import networkx
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
12 import swalign
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
13
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
14 VERBOSE = (logging.DEBUG+logging.INFO)//2
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
15 ARG_DEFAULTS = {'sam':sys.stdin, 'qual':20, 'pos':2, 'dist':1, 'choose_by':'reads', 'output':True,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
16 'visualize':0, 'viz_format':'png', 'log':sys.stderr, 'volume':logging.WARNING}
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
17 USAGE = "%(prog)s [options]"
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
18 DESCRIPTION = """Correct barcodes using an alignment of all barcodes to themselves. Reads the
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
19 alignment in SAM format and corrects the barcodes in an input "families" file (the output of
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
20 make-barcodes.awk). It will print the "families" file to stdout with barcodes (and orders)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
21 corrected."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
22
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
23
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
24 def main(argv):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
25
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
26 parser = argparse.ArgumentParser(description=DESCRIPTION)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
27 parser.set_defaults(**ARG_DEFAULTS)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
28
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
29 parser.add_argument('families', type=open_as_text_or_gzip,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
30 help='The sorted output of make-barcodes.awk. The important part is that it\'s a tab-delimited '
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
31 'file with at least 2 columns: the barcode sequence and order, and it must be sorted in '
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
32 'the same order as the "reads" in the SAM file.')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
33 parser.add_argument('reads', type=open_as_text_or_gzip,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
34 help='The fasta/q file given to the aligner. Used to get barcode sequences from read names.')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
35 parser.add_argument('sam', type=argparse.FileType('r'), nargs='?',
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
36 help='Barcode alignment, in SAM format. Omit to read from stdin. The read names must be '
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
37 'integers, representing the (1-based) order they appear in the families file.')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
38 parser.add_argument('-P', '--prepend', action='store_true',
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
39 help='Prepend the corrected barcodes and orders to the original columns.')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
40 parser.add_argument('-d', '--dist', type=int,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
41 help='NM edit distance threshold. Default: %(default)s')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
42 parser.add_argument('-m', '--mapq', type=int,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
43 help='MAPQ threshold. Default: %(default)s')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
44 parser.add_argument('-p', '--pos', type=int,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
45 help='POS tolerance. Alignments will be ignored if abs(POS - 1) is greater than this value. '
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
46 'Set to greater than the barcode length for no threshold. Default: %(default)s')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
47 parser.add_argument('-t', '--tag-len', type=int,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
48 help='Length of each half of the barcode. If not given, it will be determined from the first '
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
49 'barcode in the families file.')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
50 parser.add_argument('-c', '--choose-by', choices=('reads', 'connectivity'))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
51 parser.add_argument('--limit', type=int,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
52 help='Limit the number of lines that will be read from each input file, for testing purposes.')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
53 parser.add_argument('-S', '--structures', action='store_true',
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
54 help='Print a list of the unique isoforms')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
55 parser.add_argument('--struct-human', action='store_true')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
56 parser.add_argument('-V', '--visualize', nargs='?',
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
57 help='Produce a visualization of the unique structures write the image to this file. '
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
58 'If you omit a filename, it will be displayed in a window.')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
59 parser.add_argument('-F', '--viz-format', choices=('dot', 'graphviz', 'png'))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
60 parser.add_argument('-n', '--no-output', dest='output', action='store_false')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
61 parser.add_argument('-l', '--log', type=argparse.FileType('w'),
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
62 help='Print log messages to this file instead of to stderr. Warning: Will overwrite the file.')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
63 parser.add_argument('-q', '--quiet', dest='volume', action='store_const', const=logging.CRITICAL)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
64 parser.add_argument('-i', '--info', dest='volume', action='store_const', const=logging.INFO)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
65 parser.add_argument('-v', '--verbose', dest='volume', action='store_const', const=VERBOSE)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
66 parser.add_argument('-D', '--debug', dest='volume', action='store_const', const=logging.DEBUG,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
67 help='Print debug messages (very verbose).')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
68
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
69 args = parser.parse_args(argv[1:])
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
70
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
71 logging.basicConfig(stream=args.log, level=args.volume, format='%(message)s')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
72 tone_down_logger()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
73
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
74 logging.info('Reading the fasta/q to map read names to barcodes..')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
75 names_to_barcodes = map_names_to_barcodes(args.reads, args.limit)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
76
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
77 logging.info('Reading the SAM to build the graph of barcode relationships..')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
78 graph, reversed_barcodes = read_alignments(args.sam, names_to_barcodes, args.pos, args.mapq,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
79 args.dist, args.limit)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
80 logging.info('{} reversed barcodes'.format(len(reversed_barcodes)))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
81
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
82 logging.info('Reading the families.tsv to get the counts of each family..')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
83 family_counts = get_family_counts(args.families, args.limit)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
84
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
85 if args.structures:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
86 logging.info('Counting the unique barcode networks..')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
87 structures = count_structures(graph, family_counts)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
88 print_structures(structures, args.struct_human)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
89 if args.visualize != 0:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
90 logging.info('Generating a visualization of barcode networks..')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
91 visualize([s['graph'] for s in structures], args.visualize, args.viz_format)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
92
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
93 logging.info('Building the correction table from the graph..')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
94 corrections = make_correction_table(graph, family_counts, args.choose_by)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
95
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
96 logging.info('Reading the families.tsv again to print corrected output..')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
97 families = open_as_text_or_gzip(args.families.name)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
98 print_corrected_output(families, corrections, reversed_barcodes, args.prepend, args.limit,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
99 args.output)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
100
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
101 max_mem = resource.getrusage(resource.RUSAGE_SELF).ru_maxrss/1024
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
102 logging.info('Max memory usage: {:0.2f}MB'.format(max_mem))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
103
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
104
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
105 def detect_format(reads_file, max_lines=7):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
106 """Detect whether a file is a fastq or a fasta, based on its content."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
107 fasta_votes = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
108 fastq_votes = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
109 line_num = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
110 for line in reads_file:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
111 line_num += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
112 if line_num % 4 == 1:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
113 if line.startswith('@'):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
114 fastq_votes += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
115 elif line.startswith('>'):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
116 fasta_votes += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
117 elif line_num % 4 == 3:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
118 if line.startswith('+'):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
119 fastq_votes += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
120 elif line.startswith('>'):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
121 fasta_votes += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
122 if line_num >= max_lines:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
123 break
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
124 reads_file.seek(0)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
125 if fasta_votes > fastq_votes:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
126 return 'fasta'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
127 elif fastq_votes > fasta_votes:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
128 return 'fastq'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
129 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
130 return None
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
131
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
132
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
133 def read_fastaq(reads_file):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
134 filename = reads_file.name
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
135 if filename.endswith('.fa') or filename.endswith('.fasta'):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
136 format = 'fasta'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
137 elif filename.endswith('.fq') or filename.endswith('.fastq'):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
138 format = 'fastq'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
139 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
140 format = detect_format(reads_file)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
141 if format == 'fasta':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
142 return read_fasta(reads_file)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
143 elif format == 'fastq':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
144 return read_fastq(reads_file)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
145
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
146
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
147 def read_fasta(reads_file):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
148 """Read a FASTA file, yielding read names and sequences.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
149 NOTE: This assumes sequences are only one line!"""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
150 line_num = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
151 for line_raw in reads_file:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
152 line = line_raw.rstrip('\r\n')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
153 line_num += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
154 if line_num % 2 == 1:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
155 assert line.startswith('>'), line
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
156 read_name = line[1:]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
157 elif line_num % 2 == 0:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
158 read_seq = line
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
159 yield read_name, read_seq
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
160
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
161
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
162 def read_fastq(reads_file):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
163 """Read a FASTQ file, yielding read names and sequences.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
164 NOTE: This assumes sequences are only one line!"""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
165 line_num = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
166 for line in reads_file:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
167 line_num += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
168 if line_num % 4 == 1:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
169 assert line.startswith('@'), line
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
170 read_name = line[1:].rstrip('\r\n')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
171 elif line_num % 4 == 2:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
172 read_seq = line.rstrip('\r\n')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
173 yield read_name, read_seq
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
174
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
175
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
176 def map_names_to_barcodes(reads_file, limit=None):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
177 """Map barcode names to their sequences."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
178 names_to_barcodes = {}
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
179 read_num = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
180 for read_name, read_seq in read_fastaq(reads_file):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
181 read_num += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
182 if limit is not None and read_num > limit:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
183 break
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
184 try:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
185 name = int(read_name)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
186 except ValueError:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
187 logging.critical('non-int read name "{}"'.format(name))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
188 raise
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
189 names_to_barcodes[name] = read_seq
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
190 reads_file.close()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
191 return names_to_barcodes
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
192
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
193
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
194 def parse_alignment(sam_file, pos_thres, mapq_thres, dist_thres):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
195 """Parse the SAM file and yield reads that pass the filters.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
196 Returns (qname, rname, reversed)."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
197 line_num = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
198 for line in sam_file:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
199 line_num += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
200 if line.startswith('@'):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
201 logging.debug('Header line ({})'.format(line_num))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
202 continue
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
203 fields = line.split('\t')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
204 logging.debug('read {} -> ref {} (read seq {}):'.format(fields[2], fields[0], fields[9]))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
205 qname_str = fields[0]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
206 rname_str = fields[2]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
207 rname_fields = rname_str.split(':')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
208 if len(rname_fields) == 2 and rname_fields[1] == 'rev':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
209 reversed = True
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
210 rname_str = rname_fields[0]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
211 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
212 reversed = False
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
213 try:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
214 qname = int(qname_str)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
215 rname = int(rname_str)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
216 except ValueError:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
217 if fields[2] == '*':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
218 logging.debug('\tRead unmapped (reference == "*")')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
219 continue
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
220 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
221 logging.error('Non-integer read name(s) on line {}: "{}", "{}".'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
222 .format(line_num, qname, rname))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
223 raise
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
224 # Apply alignment quality filters.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
225 try:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
226 flags = int(fields[1])
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
227 pos = int(fields[3])
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
228 mapq = int(fields[4])
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
229 except ValueError:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
230 logging.warn('\tNon-integer flag ({}), pos ({}), or mapq ({})'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
231 .format(fields[1], fields[3], fields[4]))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
232 continue
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
233 if flags & 4:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
234 logging.debug('\tRead unmapped (flag & 4 == True)')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
235 continue
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
236 if abs(pos - 1) > pos_thres:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
237 logging.debug('\tAlignment failed pos filter: abs({} - 1) > {}'.format(pos, pos_thres))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
238 continue
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
239 if mapq < mapq_thres:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
240 logging.debug('\tAlignment failed mapq filter: {} > {}'.format(mapq, mapq_thres))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
241 continue
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
242 nm = None
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
243 for tag in fields[11:]:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
244 if tag.startswith('NM:i:'):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
245 try:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
246 nm = int(tag[5:])
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
247 except ValueError:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
248 logging.error('Invalid NM tag "{}" on line {}.'.format(tag, line_num))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
249 raise
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
250 break
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
251 assert nm is not None, line_num
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
252 if nm > dist_thres:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
253 logging.debug('\tAlignment failed NM distance filter: {} > {}'.format(nm, dist_thres))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
254 continue
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
255 yield qname, rname, reversed
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
256 sam_file.close()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
257
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
258
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
259 def read_alignments(sam_file, names_to_barcodes, pos_thres, mapq_thres, dist_thres, limit=None):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
260 """Read the alignments from the SAM file.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
261 Returns a dict mapping each reference sequence (RNAME) to sets of sequences (QNAMEs) that align to
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
262 it."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
263 graph = networkx.Graph()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
264 # This is the set of all barcodes which are involved in an alignment where the target is reversed.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
265 # Whether it's a query or reference sequence in the alignment, it's marked here.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
266 reversed_barcodes = set()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
267 # Maps correct barcode numbers to sets of original barcodes (includes correct ones).
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
268 line_num = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
269 for qname, rname, reversed in parse_alignment(sam_file, pos_thres, mapq_thres, dist_thres):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
270 line_num += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
271 if limit is not None and line_num > limit:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
272 break
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
273 # Skip self-alignments.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
274 if rname == qname:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
275 continue
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
276 rseq = names_to_barcodes[rname]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
277 qseq = names_to_barcodes[qname]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
278 # Is this an alignment to a reversed barcode?
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
279 if reversed:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
280 reversed_barcodes.add(rseq)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
281 reversed_barcodes.add(qseq)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
282 graph.add_node(rseq)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
283 graph.add_node(qseq)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
284 graph.add_edge(rseq, qseq)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
285 return graph, reversed_barcodes
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
286
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
287
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
288 def get_family_counts(families_file, limit=None):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
289 """For each family (barcode), count how many read pairs exist for each strand (order)."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
290 family_counts = {}
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
291 last_barcode = None
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
292 this_family_counts = None
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
293 line_num = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
294 for line in families_file:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
295 line_num += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
296 if limit is not None and line_num > limit:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
297 break
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
298 fields = line.rstrip('\r\n').split('\t')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
299 barcode = fields[0]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
300 order = fields[1]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
301 if barcode != last_barcode:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
302 if this_family_counts:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
303 this_family_counts['all'] = this_family_counts['ab'] + this_family_counts['ba']
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
304 family_counts[last_barcode] = this_family_counts
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
305 this_family_counts = {'ab':0, 'ba':0}
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
306 last_barcode = barcode
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
307 this_family_counts[order] += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
308 this_family_counts['all'] = this_family_counts['ab'] + this_family_counts['ba']
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
309 family_counts[last_barcode] = this_family_counts
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
310 families_file.close()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
311 return family_counts
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
312
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
313
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
314 def make_correction_table(meta_graph, family_counts, choose_by='reads'):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
315 """Make a table mapping original barcode sequences to correct barcodes.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
316 Assumes the most connected node in the graph as the correct barcode."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
317 corrections = {}
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
318 for graph in networkx.connected_component_subgraphs(meta_graph):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
319 if choose_by == 'reads':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
320 def key(bar):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
321 return family_counts[bar]['all']
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
322 elif choose_by == 'connectivity':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
323 degrees = graph.degree()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
324 def key(bar):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
325 return degrees[bar]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
326 barcodes = sorted(graph.nodes(), key=key, reverse=True)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
327 correct = barcodes[0]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
328 for barcode in barcodes:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
329 if barcode != correct:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
330 logging.debug('Correcting {} ->\n {}\n'.format(barcode, correct))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
331 corrections[barcode] = correct
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
332 return corrections
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
333
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
334
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
335 def print_corrected_output(families_file, corrections, reversed_barcodes, prepend=False, limit=None,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
336 output=True):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
337 line_num = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
338 barcode_num = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
339 barcode_last = None
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
340 corrected = {'reads':0, 'barcodes':0, 'reversed':0}
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
341 reads = [0, 0]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
342 corrections_in_this_family = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
343 for line in families_file:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
344 line_num += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
345 if limit is not None and line_num > limit:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
346 break
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
347 fields = line.rstrip('\r\n').split('\t')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
348 raw_barcode = fields[0]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
349 order = fields[1]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
350 if raw_barcode != barcode_last:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
351 # We just started a new family.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
352 barcode_num += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
353 family_info = '{}\t{}\t{}'.format(barcode_last, reads[0], reads[1])
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
354 if corrections_in_this_family:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
355 corrected['reads'] += corrections_in_this_family
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
356 corrected['barcodes'] += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
357 family_info += '\tCORRECTED!'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
358 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
359 family_info += '\tuncorrected'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
360 logging.log(VERBOSE, family_info)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
361 reads = [0, 0]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
362 corrections_in_this_family = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
363 barcode_last = raw_barcode
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
364 if order == 'ab':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
365 reads[0] += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
366 elif order == 'ba':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
367 reads[1] += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
368 if raw_barcode in corrections:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
369 correct_barcode = corrections[raw_barcode]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
370 corrections_in_this_family += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
371 # Check if the order of the barcode reverses in the correct version.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
372 # First, we check in reversed_barcodes whether either barcode was involved in a reversed
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
373 # alignment, to save time (is_alignment_reversed() does a full smith-waterman alignment).
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
374 if ((raw_barcode in reversed_barcodes or correct_barcode in reversed_barcodes) and
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
375 is_alignment_reversed(raw_barcode, correct_barcode)):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
376 # If so, then switch the order field.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
377 corrected['reversed'] += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
378 if order == 'ab':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
379 fields[1] = 'ba'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
380 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
381 fields[1] = 'ab'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
382 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
383 correct_barcode = raw_barcode
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
384 if prepend:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
385 fields.insert(0, correct_barcode)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
386 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
387 fields[0] = correct_barcode
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
388 if output:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
389 print(*fields, sep='\t')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
390 families_file.close()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
391 if corrections_in_this_family:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
392 corrected['reads'] += corrections_in_this_family
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
393 corrected['barcodes'] += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
394 logging.info('Corrected {barcodes} barcodes on {reads} read pairs, with {reversed} reversed.'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
395 .format(**corrected))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
396
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
397
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
398 def is_alignment_reversed(barcode1, barcode2):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
399 """Return True if the barcodes are reversed with respect to each other, False otherwise.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
400 "reversed" in this case meaning the alpha + beta halves are swapped.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
401 Determine by aligning the two to each other, once in their original forms, and once with the
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
402 second barcode reversed. If the smith-waterman score is higher in the reversed form, return True.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
403 """
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
404 half = len(barcode2)//2
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
405 barcode2_rev = barcode2[half:] + barcode2[:half]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
406 fwd_align = swalign.smith_waterman(barcode1, barcode2)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
407 rev_align = swalign.smith_waterman(barcode1, barcode2_rev)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
408 if rev_align.score > fwd_align.score:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
409 return True
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
410 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
411 return False
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
412
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
413
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
414 def count_structures(meta_graph, family_counts):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
415 """Count the number of unique (isomorphic) subgraphs in the main graph."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
416 structures = []
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
417 for graph in networkx.connected_component_subgraphs(meta_graph):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
418 match = False
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
419 for structure in structures:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
420 archetype = structure['graph']
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
421 if networkx.is_isomorphic(graph, archetype):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
422 match = True
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
423 structure['count'] += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
424 structure['central'] += int(is_centralized(graph, family_counts))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
425 break
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
426 if not match:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
427 size = len(graph)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
428 central = is_centralized(graph, family_counts)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
429 structures.append({'graph':graph, 'size':size, 'count':1, 'central':int(central)})
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
430 return structures
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
431
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
432
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
433 def is_centralized(graph, family_counts):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
434 """Checks if the graph is centralized in terms of where the reads are located.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
435 In a centralized graph, the node with the highest degree is the only one which (may) have more
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
436 than one read pair associated with that barcode.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
437 This returns True if that's the case, False otherwise."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
438 if len(graph) == 2:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
439 # Special-case graphs with 2 nodes, since the other algorithm doesn't work for them.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
440 # - When both nodes have a degree of 1, sorting by degree doesn't work and can result in the
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
441 # barcode with more read pairs coming second.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
442 barcode1, barcode2 = graph.nodes()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
443 counts1 = family_counts[barcode1]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
444 counts2 = family_counts[barcode2]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
445 total1 = counts1['all']
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
446 total2 = counts2['all']
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
447 logging.debug('{}: {:3d} ({}/{})\n{}: {:3d} ({}/{})\n'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
448 .format(barcode1, total1, counts1['ab'], counts1['ba'],
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
449 barcode2, total2, counts2['ab'], counts2['ba']))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
450 if (total1 >= 1 and total2 == 1) or (total1 == 1 and total2 >= 1):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
451 return True
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
452 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
453 return False
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
454 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
455 degrees = graph.degree()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
456 first = True
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
457 for barcode in sorted(graph.nodes(), key=lambda barcode: degrees[barcode], reverse=True):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
458 if not first:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
459 counts = family_counts[barcode]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
460 # How many read pairs are associated with this barcode (how many times did we see this barcode)?
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
461 try:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
462 if counts['all'] > 1:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
463 return False
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
464 except TypeError:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
465 logging.critical('barcode: {}, counts: {}'.format(barcode, counts))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
466 raise
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
467 first = False
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
468 return True
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
469
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
470
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
471 def print_structures(structures, human=True):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
472 # Define a cmp function to sort the list of structures in ascending order of size, but then
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
473 # descending order of count.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
474 def cmp_fxn(structure1, structure2):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
475 if structure1['size'] == structure2['size']:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
476 return structure2['count'] - structure1['count']
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
477 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
478 return structure1['size'] - structure2['size']
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
479 width = None
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
480 last_size = None
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
481 for structure in sorted(structures, cmp=cmp_fxn):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
482 size = structure['size']
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
483 graph = structure['graph']
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
484 if size == last_size:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
485 i += 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
486 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
487 i = 0
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
488 if width is None:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
489 width = str(len(str(structure['count'])))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
490 letters = num_to_letters(i)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
491 degrees = sorted(graph.degree().values(), reverse=True)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
492 if human:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
493 degrees_str = ' '.join(map(str, degrees))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
494 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
495 degrees_str = ','.join(map(str, degrees))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
496 if human:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
497 format_str = '{:2d}{:<3s} {count:<'+width+'d} {central:<'+width+'d} {}'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
498 print(format_str.format(size, letters+':', degrees_str, **structure))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
499 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
500 print(size, letters, structure['count'], structure['central'], degrees_str, sep='\t')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
501 last_size = size
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
502
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
503
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
504 def num_to_letters(i):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
505 """Translate numbers to letters, e.g. 1 -> A, 10 -> J, 100 -> CV"""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
506 letters = ''
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
507 while i > 0:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
508 n = (i-1) % 26
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
509 i = i // 26
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
510 if n == 25:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
511 i -= 1
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
512 letters = chr(65+n) + letters
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
513 return letters
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
514
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
515
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
516 def visualize(graphs, viz_path, args_viz_format):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
517 import matplotlib
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
518 from networkx.drawing.nx_agraph import graphviz_layout
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
519 meta_graph = networkx.Graph()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
520 for graph in graphs:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
521 add_graph(meta_graph, graph)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
522 pos = graphviz_layout(meta_graph)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
523 networkx.draw(meta_graph, pos)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
524 if viz_path:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
525 ext = os.path.splitext(viz_path)[1]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
526 if ext == '.dot':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
527 viz_format = 'graphviz'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
528 elif ext == '.png':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
529 viz_format = 'png'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
530 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
531 viz_format = args_viz_format
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
532 if viz_format == 'graphviz':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
533 from networkx.drawing.nx_pydot import write_dot
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
534 assert viz_path is not None, 'Must provide a filename to --visualize if using --viz-format "graphviz".'
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
535 base_path = os.path.splitext(viz_path)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
536 write_dot(meta_graph, base_path+'.dot')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
537 run_command('dot', '-T', 'png', '-o', base_path+'.png', base_path+'.dot')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
538 logging.info('Wrote image of graph to '+base_path+'.dot')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
539 elif viz_format == 'png':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
540 if viz_path is None:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
541 matplotlib.pyplot.show()
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
542 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
543 matplotlib.pyplot.savefig(viz_path)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
544
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
545
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
546 def add_graph(graph, subgraph):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
547 # I'm sure there's a function in the library for this, but just cause I need it quick..
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
548 for node in subgraph.nodes():
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
549 graph.add_node(node)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
550 for edge in subgraph.edges():
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
551 graph.add_edge(*edge)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
552 return graph
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
553
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
554
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
555 def open_as_text_or_gzip(path):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
556 """Return an open file-like object reading the path as a text file or a gzip file, depending on
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
557 which it looks like."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
558 if detect_gzip(path):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
559 return gzip.open(path, 'r')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
560 else:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
561 return open(path, 'rU')
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
562
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
563
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
564 def detect_gzip(path):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
565 """Return True if the file looks like a gzip file: ends with .gz or contains non-ASCII bytes."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
566 ext = os.path.splitext(path)[1]
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
567 if ext == '.gz':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
568 return True
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
569 elif ext in ('.txt', '.tsv', '.csv'):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
570 return False
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
571 with open(path) as fh:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
572 is_not_ascii = detect_non_ascii(fh.read(100))
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
573 if is_not_ascii:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
574 return True
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
575
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
576
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
577 def detect_non_ascii(bytes, max_test=100):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
578 """Return True if any of the first "max_test" bytes are non-ASCII (the high bit set to 1).
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
579 Return False otherwise."""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
580 for i, char in enumerate(bytes):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
581 # Is the high bit a 1?
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
582 if ord(char) & 128:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
583 return True
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
584 if i >= max_test:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
585 return False
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
586 return False
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
587
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
588
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
589 def run_command(*command):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
590 try:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
591 exit_status = subprocess.call(command)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
592 except subprocess.CalledProcessError as cpe:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
593 exit_status = cpe.returncode
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
594 except OSError:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
595 exit_status = None
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
596 return exit_status
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
597
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
598
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
599 def tone_down_logger():
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
600 """Change the logging level names from all-caps to capitalized lowercase.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
601 E.g. "WARNING" -> "Warning" (turn down the volume a bit in your log files)"""
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
602 for level in (logging.CRITICAL, logging.ERROR, logging.WARNING, logging.INFO, logging.DEBUG):
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
603 level_name = logging.getLevelName(level)
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
604 logging.addLevelName(level, level_name.capitalize())
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
605
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
606
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
607 if __name__ == '__main__':
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
608 sys.exit(main(sys.argv))