Mercurial > repos > artbio > repenrich

annotate RepEnrich.py @ 12:89e05f831259 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
author artbio
date Mon, 18 Mar 2024 09:39:44 +0000
parents 6f4143893463
children 530626b0757c
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
1 import argparse
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
2 import csv
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
3 import os
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
4 import shlex
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
5 import subprocess
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
6 import sys
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
7
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
8 import numpy
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
9
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
10
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
11 parser = argparse.ArgumentParser(description='''
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
12 Repenrich aligns reads to Repeat Elements pseudogenomes\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
13 and counts aligned reads. RepEnrich_setup must be run\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
14 before its use''')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
15 parser.add_argument('--annotation_file', action='store',
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
16 metavar='annotation_file',
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
17 help='RepeatMasker.org annotation file for your\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
18 organism. The file may be downloaded from\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
19 RepeatMasker.org. E.g. hg19_repeatmasker.txt')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
20 parser.add_argument('--outputfolder', action='store', metavar='outputfolder',
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
21 help='Folder that will contain results. Should be the\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
22 same as the one used for RepEnrich_setup.\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
23 Example: ./outputfolder')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
24 parser.add_argument('--outputprefix', action='store', metavar='outputprefix',
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
25 help='Prefix name for Repenrich output files.')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
26 parser.add_argument('--setup_folder', action='store', metavar='setup_folder',
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
27 help='Folder produced by RepEnrich_setup which contains\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
28 repeat element pseudogenomes.')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
29 parser.add_argument('--fastqfile', action='store', metavar='fastqfile',
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
30 help='File of fastq reads mapping to multiple\
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
31 locations. Example: /data/multimap.fastq')
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
32 parser.add_argument('--alignment_bam', action='store', metavar='alignment_bam',
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
33 help='Bam alignments of unique mapper reads.')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
34 parser.add_argument('--pairedend', action='store', dest='pairedend',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
35 default='FALSE',
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
36 help='Change to TRUE for paired-end fastq files.\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
37 Default FALSE')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
38 parser.add_argument('--fastqfile2', action='store', dest='fastqfile2',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
39 metavar='fastqfile2', default='none',
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
40 help='fastqfile #2 when using paired-end option.\
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
41 Default none')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
42 parser.add_argument('--cpus', action='store', dest='cpus', metavar='cpus',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
43 default="1", type=int,
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
44 help='Number of CPUs. The more cpus the\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
45 faster RepEnrich performs. Default: "1"')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
46
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
47 args = parser.parse_args()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
48
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
49 # parameters
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
50 annotation_file = args.annotation_file
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
51 outputfolder = args.outputfolder
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
52 outputfile_prefix = args.outputprefix
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
53 setup_folder = args.setup_folder
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
54 repeat_bed = os.path.join(setup_folder, 'repnames.bed')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
55 unique_mapper_bam = args.alignment_bam
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
56 fastqfile_1 = args.fastqfile
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
57 fastqfile_2 = args.fastqfile2
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
58 cpus = args.cpus
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
59 b_opt = "-k1 -p 1 --quiet"
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
60 # Change if simple repeats are differently annotated in your organism
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
61 simple_repeat = "Simple_repeat"
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
62 paired_end = args.pairedend
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
63
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
64 # check that the programs we need are available
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
65 try:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
66 subprocess.call(shlex.split("coverageBed -h"),
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
67 stdout=open(os.devnull, 'wb'),
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
68 stderr=open(os.devnull, 'wb'))
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
69 subprocess.call(shlex.split("bowtie --version"),
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
70 stdout=open(os.devnull, 'wb'),
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
71 stderr=open(os.devnull, 'wb'))
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
72 except OSError:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
73 print("Error: Bowtie or bedtools not loaded")
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
74 raise
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
75
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
76 # define a csv reader that reads space deliminated files
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
77 print('Preparing for analysis using RepEnrich...')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
78 csv.field_size_limit(sys.maxsize)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
79
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
80
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
81 def import_text(filename, separator):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
82 for line in csv.reader(open(filename), delimiter=separator,
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
83 skipinitialspace=True):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
84 if line:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
85 yield line
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
86
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
87
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
88 # build dictionaries to convert repclass and rep families
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
89 repeatclass, repeatfamily = {}, {}
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
90 repeats = import_text(annotation_file, ' ')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
91 # skip three first lines of the iterator
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
92 for line in range(3):
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
93 next(repeats)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
94 for repeat in repeats:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
95 classfamily = []
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
96 classfamily = repeat[10].split('/')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
97 matching_repeat = repeat[9].translate(str.maketrans('()/', '___'))
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
98 repeatclass[matching_repeat] = classfamily[0]
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
99 if len(classfamily) == 2:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
100 repeatfamily[matching_repeat] = classfamily[1]
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
101 else:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
102 repeatfamily[matching_repeat] = classfamily[0]
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
103
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
104 # build list of repeats initializing dictionaries for downstream analysis'
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
105 repgenome_path = os.path.join(setup_folder, 'repgenomes_key.txt')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
106 reptotalcounts = {line[0]: 0 for line in import_text(repgenome_path, '\t')}
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
107 fractionalcounts = {line[0]: 0 for line in import_text(repgenome_path, '\t')}
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
108 classtotalcounts = {repeatclass[line[0]]: 0 for line in import_text(
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
109 repgenome_path, '\t') if line[0] in repeatclass}
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
110 classfractionalcounts = {repeatclass[line[0]]: 0 for line in import_text(
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
111 repgenome_path, '\t') if line[0] in repeatclass}
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
112 familytotalcounts = {repeatfamily[line[0]]: 0 for line in import_text(
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
113 repgenome_path, '\t') if line[0] in repeatfamily}
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
114 familyfractionalcounts = {repeatfamily[line[0]]: 0 for line in import_text(
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
115 repgenome_path, '\t') if line[0] in repeatfamily}
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
116 reptotalcounts_simple = {(simple_repeat if line[0] in repeatfamily and
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
117 repeatfamily[line[0]] == simple_repeat else
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
118 line[0]): 0 for line in import_text(
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
119 repgenome_path, '\t')}
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
120 repeat_key = {line[0]: int(line[1]) for line in import_text(
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
121 repgenome_path, '\t')}
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
122 rev_repeat_key = {int(line[1]): line[0] for line in import_text(
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
123 repgenome_path, '\t')}
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
124 repeat_list = [line[0] for line in import_text(repgenome_path, '\t')]
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
125
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
126 # map the repeats to the psuedogenomes:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
127 if not os.path.exists(outputfolder):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
128 os.mkdir(outputfolder)
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
129
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
130 # Conduct the regions sorting
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
131 fileout = os.path.join(outputfolder, f"{outputfile_prefix}_regionsorter.txt")
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
132 command = shlex.split(f"coverageBed -abam {unique_mapper_bam} -b \
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
133 {os.path.join(setup_folder, 'repnames.bed')}")
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
134 with open(fileout, 'w') as stdout:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
135 subprocess.run(command, stdout=stdout, check=True)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
136
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
137 counts = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
138 sumofrepeatreads = 0
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
139 with open(fileout) as filein:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
140 for line in filein:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
141 line = line.split('\t')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
142 if not str(repeat_key[line[3]]) in counts:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
143 counts[str(repeat_key[line[3]])] = 0
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
144 counts[str(repeat_key[line[3]])] += int(line[4])
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
145 sumofrepeatreads += int(line[4])
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
146 print(f"Identified {sumofrepeatreads} unique reads that \
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
147 mapped to repeats.")
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
148
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
149
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
150 def run_bowtie(metagenome, fastqfile, folder):
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
151 metagenomepath = os.path.join(setup_folder, metagenome)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
152 output_file = os.path.join(folder, f"{metagenome}.bowtie")
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
153 command = shlex.split(f"bowtie {b_opt} {metagenomepath} {fastqfile}")
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
154 with open(output_file, 'w') as stdout:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
155 return subprocess.Popen(command, stdout=stdout)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
156
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
157
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
158 if paired_end == 'FALSE':
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
159 folder_pair1 = os.path.join(outputfolder, 'pair1_bowtie')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
160 os.makedirs(folder_pair1, exist_ok=True)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
161
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
162 print("Processing repeat pseudogenomes...")
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
163 processes = []
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
164 ticker = 0
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
165
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
166 for metagenome in repeat_list:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
167 processes.append(run_bowtie(metagenome, fastqfile_1, folder_pair1))
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
168 ticker += 1
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
169 if ticker == cpus:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
170 for p in processes:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
171 p.communicate()
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
172 ticker = 0
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
173 processes = []
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
174
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
175 for p in processes:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
176 p.communicate()
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
177 # Combine the output from both read pairs:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
178 print('Sorting and combining the output for both read pairs....')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
179 sorted_bowtie = os.path.join(outputfolder, 'sorted_bowtie')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
180 os.makedirs(sorted_bowtie, exist_ok=True)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
181 for metagenome in repeat_list:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
182 file1 = os.path.join(folder_pair1, f"{metagenome}.bowtie")
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
183 fileout = os.path.join(sorted_bowtie, f"{metagenome}.bowtie")
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
184 with open(fileout, 'w') as stdout:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
185 p1 = subprocess.Popen(['cat', file1], stdout=subprocess.PIPE)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
186 p2 = subprocess.Popen(['cut', '-f1'], stdin=p1.stdout,
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
187 stdout=subprocess.PIPE)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
188 p3 = subprocess.Popen(['cut', '-f1', "-d/"], stdin=p2.stdout,
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
189 stdout=subprocess.PIPE)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
190 p4 = subprocess.Popen(['sort'], stdin=p3.stdout,
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
191 stdout=subprocess.PIPE)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
192 p5 = subprocess.Popen(['uniq'], stdin=p4.stdout, stdout=stdout)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
193 p5.communicate()
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
194 stdout.close()
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
195 print('completed ...')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
196 else:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
197 folder_pair1 = os.path.join(outputfolder, 'pair1_bowtie')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
198 folder_pair2 = os.path.join(outputfolder, 'pair2_bowtie')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
199 os.makedirs(folder_pair1, exist_ok=True)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
200 os.makedirs(folder_pair2, exist_ok=True)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
201
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
202 print("Processing repeat pseudogenomes...")
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
203 ps, psb, ticker = [], [], 0
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
204
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
205 for metagenome in repeat_list:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
206 ps.append(run_bowtie(metagenome, fastqfile_1, folder_pair1))
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
207 ticker += 1
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
208 if fastqfile_2 != 'none':
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
209 psb.append(run_bowtie(metagenome, fastqfile_2, folder_pair2))
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
210 ticker += 1
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
211 if ticker >= cpus:
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
212 for p in ps:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
213 p.communicate()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
214 for p in psb:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
215 p.communicate()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
216 ticker = 0
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
217 ps = []
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
218 psb = []
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
219
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
220 for p in ps:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
221 p.communicate()
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
222 for p in psb:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
223 p.communicate()
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
224 # combine the output from both read pairs:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
225 print('Sorting and combining the output for both read pairs...')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
226 if not os.path.exists(outputfolder + os.path.sep + 'sorted_bowtie'):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
227 os.mkdir(outputfolder + os.path.sep + 'sorted_bowtie')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
228 sorted_bowtie = outputfolder + os.path.sep + 'sorted_bowtie'
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
229 for metagenome in repeat_list:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
230 file1 = folder_pair1 + os.path.sep + metagenome + '.bowtie'
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
231 file2 = folder_pair2 + os.path.sep + metagenome + '.bowtie'
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
232 fileout = sorted_bowtie + os.path.sep + metagenome + '.bowtie'
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
233 with open(fileout, 'w') as stdout:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
234 p1 = subprocess.Popen(['cat', file1, file2],
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
235 stdout=subprocess.PIPE)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
236 p2 = subprocess.Popen(['cut', '-f1', "-d "], stdin=p1.stdout,
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
237 stdout=subprocess.PIPE)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
238 p3 = subprocess.Popen(['cut', '-f1', "-d/"], stdin=p2.stdout,
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
239 stdout=subprocess.PIPE)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
240 p4 = subprocess.Popen(['sort'], stdin=p3.stdout,
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
241 stdout=subprocess.PIPE)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
242 p5 = subprocess.Popen(['uniq'], stdin=p4.stdout, stdout=stdout)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
243 p5.communicate()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
244 stdout.close()
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
245 print('completed ...')
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
246
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
247 # build a file of repeat keys for all reads
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
248 print('Writing and processing intermediate files...')
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
249 sorted_bowtie = os.path.join(outputfolder, 'sorted_bowtie')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
250 sumofrepeatreads = 0
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
251 readid = {}
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
252
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
253 for rep in repeat_list:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
254 for data in import_text(
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
255 f"{os.path.join(sorted_bowtie, rep)}.bowtie", '\t'):
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
256 readid[data[0]] = ''
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
257
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
258 for rep in repeat_list:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
259 for data in import_text(
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
260 f"{os.path.join(sorted_bowtie, rep)}.bowtie", '\t'):
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
261 readid[data[0]] += f"{repeat_key[rep]},"
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
262
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
263 for subfamilies in readid.values():
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
264 if subfamilies not in counts:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
265 counts[subfamilies] = 0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
266 counts[subfamilies] += 1
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
267 sumofrepeatreads += 1
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
268
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
269 print(f'Identified {sumofrepeatreads} reads that mapped to \
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
270 repeats for unique and multimappers.')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
271 print("Conducting final calculations...")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
272
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
273 # building the total counts for repeat element enrichment...
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
274 for x in counts.keys():
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
275 count = counts[x]
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
276 x = x.strip(',').split(',')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
277 for i in x:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
278 reptotalcounts[rev_repeat_key[int(i)]] += int(count)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
279 # building the fractional counts for repeat element enrichment...
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
280 for x in counts.keys():
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
281 count = counts[x]
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
282 x = x.strip(',') .split(',')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
283 splits = len(x)
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
284 for i in x:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
285 fractionalcounts[rev_repeat_key[int(i)]] += float(
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
286 numpy.divide(float(count), float(splits)))
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
287 # building categorized table of repeat element enrichment...
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
288 repcounts = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
289 repcounts['other'] = 0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
290 for key in counts.keys():
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
291 key_list = key.strip(',').split(',')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
292 repname = ''
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
293 for i in key_list:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
294 repname = os.path.join(repname, rev_repeat_key[int(i)])
10
6f4143893463 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit e3881f05134c6f50889d0376d27e1c232251f8b3
artbio
parents: 4
diff changeset
295 repcounts[repname] = counts[key]
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
296 # building the total counts for class enrichment...
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
297 for key in reptotalcounts.keys():
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
298 classtotalcounts[repeatclass[key]] += reptotalcounts[key]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
299 # building total counts for family enrichment...
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
300 for key in reptotalcounts.keys():
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
301 familytotalcounts[repeatfamily[key]] += reptotalcounts[key]
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
302 # building unique counts table
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
303 repcounts2 = {}
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
304 for rep in repeat_list:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
305 if "/" + rep in repcounts:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
306 repcounts2[rep] = repcounts["/" + rep]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
307 else:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
308 repcounts2[rep] = 0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
309 # building the fractionalcounts counts for class enrichment...
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
310 for key in fractionalcounts.keys():
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
311 classfractionalcounts[repeatclass[key]] += fractionalcounts[key]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
312 # building fractional counts for family enrichment...
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
313 for key in fractionalcounts.keys():
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
314 familyfractionalcounts[repeatfamily[key]] += fractionalcounts[key]
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
315
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
316 # print output to file of the categorized counts and total overlapping counts:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
317 print('Writing final output...')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
318 with open(f"{os.path.join(outputfolder, outputfile_prefix)}_"
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
319 f"class_fraction_counts.txt", 'w') as fout:
2
15e3e29f310e planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit c89c33e5ea8fc63f3ea5c0f66ebc5fa822ac734b
artbio
parents: 0
diff changeset
320 for key in sorted(classfractionalcounts.keys()):
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
321 fout.write(f"{key}\t{classfractionalcounts[key]}\n")
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
322
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
323 with open(f"{os.path.join(outputfolder, outputfile_prefix)}_"
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
324 f"family_fraction_counts.txt", 'w') as fout:
2
15e3e29f310e planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit c89c33e5ea8fc63f3ea5c0f66ebc5fa822ac734b
artbio
parents: 0
diff changeset
325 for key in sorted(familyfractionalcounts.keys()):
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
326 fout.write(f"{key}\t{familyfractionalcounts[key]}\n")
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
327
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
328 with open(f"{os.path.join(outputfolder, outputfile_prefix)}_"
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
329 f"fraction_counts.txt", 'w') as fout:
2
15e3e29f310e planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit c89c33e5ea8fc63f3ea5c0f66ebc5fa822ac734b
artbio
parents: 0
diff changeset
330 for key in sorted(fractionalcounts.keys()):
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
331 fout.write(f"{key}\t{repeatclass[key]}\t{repeatfamily[key]}\t"
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 10
diff changeset
332 f"{int(fractionalcounts[key])}\n")