Mercurial > repos > artbio > repenrich

annotate RepEnrich_setup.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 6bba3e33c2e7
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 #!/usr/bin/env python
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
2 import argparse
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
3 import csv
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
4 import os
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
5 import shlex
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
6 import subprocess
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
7 import sys
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
8
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
9 from Bio import SeqIO
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
10 from Bio.Seq import Seq
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
11 from Bio.SeqRecord import SeqRecord
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
12
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
13 parser = argparse.ArgumentParser(description='''
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
14 Prepartion of repetive element pseudogenomes bowtie\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
15 indexes and annotation files used by RepEnrich.py enrichment.''',
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
16 prog='getargs_genome_maker.py')
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
17 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
18 metavar='annotation_file',
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
19 help='''Repeat masker annotation of the genome of\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
20 interest. Download from RepeatMasker.org\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
21 Example: mm9.fa.out''')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
22 parser.add_argument('--genomefasta', action='store', metavar='genomefasta',
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
23 help='''Genome of interest in fasta format.\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
24 Example: mm9.fa''')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
25 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: 11
diff changeset
26 help='''Folder that contains bowtie indexes of repeats and\
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
27 repeat element psuedogenomes.\
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
28 Example working/setup''')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
29 parser.add_argument('--gaplength', action='store', dest='gaplength',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
30 metavar='gaplength', default='200', type=int,
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
31 help='''Length of the N-spacer in the\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
32 repeat pseudogenomes. Default 200''')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
33 parser.add_argument('--flankinglength', action='store', dest='flankinglength',
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
34 metavar='flankinglength', default='25', type=int,
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
35 help='''Length of the flanking regions used to build\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
36 repeat pseudogenomes. Flanking length should be set\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
37 according to the length of your reads.\
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
38 Default 25, for 50 nt reads''')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
39 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
40
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
41 # parameters from argsparse
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
42 gapl = args.gaplength
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
43 flankingl = args.flankinglength
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
44 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
45 genomefasta = args.genomefasta
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
46 setup_folder = args.setup_folder
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
47
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
48 # 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
49 try:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
50 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
51 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
52 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
53 except OSError:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
54 print("Error: Bowtie not available in the path")
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
55 raise
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
56
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
57 # Define a text importer
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
58 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
59
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
60
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
61 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
62 for line in csv.reader(open(os.path.realpath(filename)),
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
63 delimiter=separator, skipinitialspace=True):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
64 if line:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
65 yield line
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
66
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
67
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
68 # Make a setup folder
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
69 if not os.path.exists(setup_folder):
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
70 os.makedirs(setup_folder)
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
71 # load genome into dictionary and compute length
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
72 g = SeqIO.to_dict(SeqIO.parse(genomefasta, "fasta"))
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
73 idxgenome, lgenome, genome = {}, {}, {}
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
74
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
75 for k, chr in enumerate(g.keys()):
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
76 genome[chr] = g[chr].seq
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
77 lgenome[chr] = len(genome[chr])
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
78 idxgenome[chr] = k
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 # Build a bedfile of repeatcoordinates to use by RepEnrich region_sorter
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
81 repeat_elements = []
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
82 # these dictionaries will contain lists
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
83 rep_chr, rep_start, rep_end = {}, {}, {}
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
84 fin = import_text(annotation_file, ' ')
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
85 with open(os.path.join(setup_folder, 'repnames.bed'), 'w') as fout:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
86 for i in range(3):
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
87 next(fin)
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
88 for line in fin:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
89 repname = line[9].translate(str.maketrans('()/', '___'))
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
90 if repname not in repeat_elements:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
91 repeat_elements.append(repname)
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
92 repchr = line[4]
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
93 repstart = line[5]
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
94 repend = line[6]
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
95 fout.write('\t'.join([repchr, repstart, repend, repname]) + '\n')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
96 if repname in rep_chr:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
97 rep_chr[repname].append(repchr)
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
98 rep_start[repname].append(repstart)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
99 rep_end[repname].append(repend)
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
100 else:
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
101 rep_chr[repname] = [repchr]
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
102 rep_start[repname] = [repstart]
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
103 rep_end[repname] = [repend]
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
104
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
105 # sort repeat_elements and print them in repgenomes_key.txt
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
106 with open(os.path.join(setup_folder, 'repgenomes_key.txt'), 'w') as fout:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
107 for i, repeat in enumerate(sorted(repeat_elements)):
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
108 fout.write('\t'.join([repeat, str(i)]) + '\n')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
109
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
110 # generate spacer for pseudogenomes
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
111 spacer = ''.join(['N' for i in range(gapl)])
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
112
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
113 # generate metagenomes and save them to FASTA files for bowtie build
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
114 for repname in rep_chr:
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
115 metagenome = ''
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
116 for i, repeat in enumerate(rep_chr[repname]):
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
117 try:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
118 chromosome = rep_chr[repname][i]
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
119 start = max(int(rep_start[repname][i]) - flankingl, 0)
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
120 end = min(int(rep_end[repname][i]) + flankingl,
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
121 int(lgenome[chr])-1) + 1
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
122 metagenome = f"{metagenome}{spacer}{genome[chromosome][start:end]}"
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
123 except KeyError:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
124 print("Unrecognised Chromosome: " + rep_chr[repname][i])
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
125
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
126 # Create Fasta of repeat pseudogenome
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
127 fastafilename = f"{os.path.join(setup_folder, repname)}.fa"
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
128 record = SeqRecord(Seq(metagenome), id=repname, name='', description='')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
129 SeqIO.write(record, fastafilename, "fasta")
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
130
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
131 # Generate repeat pseudogenome bowtie index
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
132 bowtie_build_cmd = ["bowtie-build", "-f", fastafilename,
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
133 os.path.join(setup_folder, repname)]
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
134 subprocess.run(bowtie_build_cmd, check=True)