annotate vsnp_statistics.py @ 7:6dc6dd4666e3 draft

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
author iuc
date Tue, 16 Nov 2021 20:10:48 +0000
parents 5a5cf6f024bf
children 40b97055bb99
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
1 #!/usr/bin/env python
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
2
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
3 import argparse
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
4 import gzip
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
5 import os
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
6 from functools import partial
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
7
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
8 import numpy
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
9 import pandas
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
10 from Bio import SeqIO
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
11
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
12
7
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
13 class Statistics:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
14
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
15 def __init__(self, reference, fastq_file, file_size, total_reads, mean_read_length, mean_read_quality, reads_passing_q30):
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
16 self.reference = reference
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
17 self.fastq_file = fastq_file
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
18 self.file_size = file_size
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
19 self.total_reads = total_reads
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
20 self.mean_read_length = mean_read_length
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
21 self.mean_read_quality = mean_read_quality
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
22 self.reads_passing_q30 = reads_passing_q30
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
23
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
24
5
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
25 def nice_size(size):
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
26 # Returns a readably formatted string with the size
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
27 words = ['bytes', 'KB', 'MB', 'GB', 'TB', 'PB', 'EB']
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
28 prefix = ''
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
29 try:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
30 size = float(size)
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
31 if size < 0:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
32 size = abs(size)
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
33 prefix = '-'
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
34 except Exception:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
35 return '??? bytes'
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
36 for ind, word in enumerate(words):
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
37 step = 1024 ** (ind + 1)
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
38 if step > size:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
39 size = size / float(1024 ** ind)
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
40 if word == 'bytes': # No decimals for bytes
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
41 return "%s%d bytes" % (prefix, size)
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
42 return "%s%.1f %s" % (prefix, size, word)
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
43 return '??? bytes'
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
44
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
45
7
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
46 def get_statistics(dbkey, fastq_file, gzipped):
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
47 sampling_size = 10000
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
48 # Read fastq_file into a data fram to
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
49 # get the phred quality scores.
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
50 _open = partial(gzip.open, mode='rt') if gzipped else open
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
51 with _open(fastq_file) as fh:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
52 identifiers = []
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
53 seqs = []
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
54 letter_annotations = []
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
55 for seq_record in SeqIO.parse(fh, "fastq"):
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
56 identifiers.append(seq_record.id)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
57 seqs.append(seq_record.seq)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
58 letter_annotations.append(seq_record.letter_annotations["phred_quality"])
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
59 # Convert lists to Pandas series.
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
60 s1 = pandas.Series(identifiers, name='id')
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
61 s2 = pandas.Series(seqs, name='seq')
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
62 # Gather Series into a data frame.
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
63 fastq_df = pandas.DataFrame(dict(id=s1, seq=s2)).set_index(['id'])
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
64 # Starting at row 3, keep every 4 row
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
65 # random sample specified number of rows.
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
66 file_size = nice_size(os.path.getsize(fastq_file))
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
67 total_reads = len(seqs)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
68 # Mean Read Length
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
69 if sampling_size > total_reads:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
70 sampling_size = total_reads
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
71 try:
5
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
72 fastq_df = fastq_df.iloc[3::4].sample(sampling_size)
7
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
73 except ValueError:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
74 fastq_df = fastq_df.iloc[3::4].sample(sampling_size, replace=True)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
75 dict_mean = {}
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
76 list_length = []
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
77 i = 0
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
78 for id, seq, in fastq_df.iterrows():
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
79 dict_mean[id] = numpy.mean(letter_annotations[i])
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
80 list_length.append(len(seq.array[0]))
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
81 i += 1
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
82 mean_read_length = '%.1f' % numpy.mean(list_length)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
83 # Mean Read Quality
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
84 df_mean = pandas.DataFrame.from_dict(dict_mean, orient='index', columns=['ave'])
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
85 mean_read_quality = '%.1f' % df_mean['ave'].mean()
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
86 # Reads Passing Q30
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
87 reads_gt_q30 = len(df_mean[df_mean['ave'] >= 30])
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
88 reads_passing_q30 = '{:10.2f}'.format(reads_gt_q30 / sampling_size)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
89 stats = Statistics(dbkey, os.path.basename(fastq_file), file_size, total_reads, mean_read_length,
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
90 mean_read_quality, reads_passing_q30)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
91 return stats
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
92
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
93
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
94 def accrue_statistics(dbkey, read1, read2, gzipped):
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
95 read1_stats = get_statistics(dbkey, read1, gzipped)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
96 if read2 is None:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
97 read2_stats = None
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
98 else:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
99 read2_stats = get_statistics(dbkey, read2, gzipped)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
100 return read1_stats, read2_stats
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
101
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
102
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
103 def output_statistics(read1_stats, read2_stats, idxstats_file, metrics_file, output_file):
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
104 paired_reads = read2_stats is not None
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
105 if paired_reads:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
106 columns = ['Read1 FASTQ', 'File Size', 'Reads', 'Mean Read Length', 'Mean Read Quality',
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
107 'Reads Passing Q30', 'Read2 FASTQ', 'File Size', 'Reads', 'Mean Read Length', 'Mean Read Quality',
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
108 'Reads Passing Q30', 'Total Reads', 'All Mapped Reads', 'Unmapped Reads',
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
109 'Unmapped Reads Percentage of Total', 'Reference with Coverage', 'Average Depth of Coverage',
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
110 'Good SNP Count', 'Reference']
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
111 else:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
112 columns = ['FASTQ', 'File Size', 'Mean Read Length', 'Mean Read Quality', 'Reads Passing Q30',
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
113 'Total Reads', 'All Mapped Reads', 'Unmapped Reads', 'Unmapped Reads Percentage of Total',
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
114 'Reference with Coverage', 'Average Depth of Coverage', 'Good SNP Count', 'Reference']
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
115 with open(output_file, "w") as outfh:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
116 # Make sure the header starts with a # so
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
117 # MultiQC can properly handle the output.
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
118 outfh.write("%s\n" % "\t".join(columns))
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
119 line_items = []
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
120 # Get the current stats and associated files.
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
121 # Get and output the statistics.
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
122 line_items.append(read1_stats.fastq_file)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
123 line_items.append(read1_stats.file_size)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
124 if paired_reads:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
125 line_items.append(read1_stats.total_reads)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
126 line_items.append(read1_stats.mean_read_length)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
127 line_items.append(read1_stats.mean_read_quality)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
128 line_items.append(read1_stats.reads_passing_q30)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
129 if paired_reads:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
130 line_items.append(read2_stats.fastq_file)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
131 line_items.append(read2_stats.file_size)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
132 line_items.append(read2_stats.total_reads)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
133 line_items.append(read2_stats.mean_read_length)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
134 line_items.append(read2_stats.mean_read_quality)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
135 line_items.append(read2_stats.reads_passing_q30)
5
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
136 # Total Reads
7
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
137 if paired_reads:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
138 total_reads = read1_stats.total_reads + read2_stats.total_reads
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
139 else:
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
140 total_reads = read1_stats.total_reads
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
141 line_items.append(total_reads)
5
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
142 # All Mapped Reads
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
143 all_mapped_reads, unmapped_reads = process_idxstats_file(idxstats_file)
7
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
144 line_items.append(all_mapped_reads)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
145 line_items.append(unmapped_reads)
5
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
146 # Unmapped Reads Percentage of Total
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
147 if unmapped_reads > 0:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
148 unmapped_reads_percentage = '{:10.2f}'.format(unmapped_reads / total_reads)
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
149 else:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
150 unmapped_reads_percentage = 0
7
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
151 line_items.append(unmapped_reads_percentage)
5
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
152 # Reference with Coverage
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
153 ref_with_coverage, avg_depth_of_coverage, good_snp_count = process_metrics_file(metrics_file)
7
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
154 line_items.append(ref_with_coverage)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
155 line_items.append(avg_depth_of_coverage)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
156 line_items.append(good_snp_count)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
157 line_items.append(read1_stats.reference)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
158 outfh.write('%s\n' % '\t'.join(str(x) for x in line_items))
5
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
159
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
160
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
161 def process_idxstats_file(idxstats_file):
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
162 all_mapped_reads = 0
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
163 unmapped_reads = 0
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
164 with open(idxstats_file, "r") as fh:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
165 for i, line in enumerate(fh):
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
166 line = line.rstrip('\r\n')
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
167 items = line.split("\t")
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
168 if i == 0:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
169 # NC_002945.4 4349904 213570 4047
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
170 all_mapped_reads = int(items[2])
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
171 elif i == 1:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
172 # * 0 0 82774
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
173 unmapped_reads = int(items[3])
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
174 return all_mapped_reads, unmapped_reads
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
175
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
176
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
177 def process_metrics_file(metrics_file):
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
178 ref_with_coverage = '0%'
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
179 avg_depth_of_coverage = 0
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
180 good_snp_count = 0
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
181 with open(metrics_file, "r") as ifh:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
182 for i, line in enumerate(ifh):
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
183 if i == 0:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
184 # Skip comments.
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
185 continue
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
186 line = line.rstrip('\r\n')
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
187 items = line.split("\t")
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
188 if i == 1:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
189 # MarkDuplicates 10.338671 98.74%
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
190 ref_with_coverage = items[3]
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
191 avg_depth_of_coverage = items[2]
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
192 elif i == 2:
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
193 # VCFfilter 611
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
194 good_snp_count = items[1]
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
195 return ref_with_coverage, avg_depth_of_coverage, good_snp_count
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
196
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
197
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
198 parser = argparse.ArgumentParser()
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
199
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
200 parser.add_argument('--dbkey', action='store', dest='dbkey', help='Reference dbkey')
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
201 parser.add_argument('--gzipped', action='store_true', dest='gzipped', required=False, default=False, help='Input files are gzipped')
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
202 parser.add_argument('--output', action='store', dest='output', help='Output Excel statistics file')
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
203 parser.add_argument('--read1', action='store', dest='read1', help='Required: single read')
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
204 parser.add_argument('--read2', action='store', dest='read2', required=False, default=None, help='Optional: paired read')
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
205 parser.add_argument('--samtools_idxstats', action='store', dest='samtools_idxstats', help='Output of samtools_idxstats')
7
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
206 parser.add_argument('--vsnp_azc_metrics', action='store', dest='vsnp_azc_metrics', help='Output of vsnp_add_zero_coverage')
5
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
207
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
208 args = parser.parse_args()
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
209
7
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
210 stats_list = []
5
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
211 idxstats_files = []
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
212 metrics_files = []
5a5cf6f024bf "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 92f46d4bb55b582f05ac3c4b094307f114cbf98f"
iuc
parents:
diff changeset
213 # Accumulate inputs.
7
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
214 read1_stats, read2_stats = accrue_statistics(args.dbkey, args.read1, args.read2, args.gzipped)
6dc6dd4666e3 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 5
diff changeset
215 output_statistics(read1_stats, read2_stats, args.samtools_idxstats, args.vsnp_azc_metrics, args.output)