Mercurial > repos > artbio > repenrich

annotate RepEnrich_setup.py @ 15:2e3d976e7d5d draft default tip

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 03183e29f807ec33548016a7c4144f52720b7b9e
author artbio
date Sun, 21 Apr 2024 09:44:51 +0000
parents bf866bedd4b4
children
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 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
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
7 from collections import defaultdict
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
8 from concurrent.futures import ProcessPoolExecutor
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
9
0
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 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
12 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
13 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
14
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
15 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
16 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
17 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
18 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
19 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
20 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
21 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
22 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
23 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
24 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
25 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
26 Example: mm9.fa''')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
27 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
28 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
29 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
30 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
31 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
32 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
33 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
34 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
35 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
36 Default 25, for 50 nt reads''')
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
37 parser.add_argument('--cpus', action='store', dest='cpus', metavar='cpus',
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
38 default="1", type=int,
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
39 help='Number of CPUs. The more cpus the\
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
40 faster RepEnrich performs. Default: "1"')
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
41 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
42
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
43 # 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
44 gapl = args.gaplength
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
45 flankingl = args.flankinglength
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
46 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
47 genomefasta = args.genomefasta
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
48 cpus = args.cpus
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
49
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
50
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
51 def starts_with_numerical(list):
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
52 try:
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
53 if len(list) == 0:
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
54 return False
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
55 int(list[0])
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
56 return True
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
57 except ValueError:
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
58 return False
0
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
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
61 # define a text importer for .out/.txt format of repbase
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
62 def import_text(filename, separator):
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
63 csv.field_size_limit(sys.maxsize)
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
64 file = csv.reader(open(filename), delimiter=separator,
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
65 skipinitialspace=True)
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
66 return [line for line in file if starts_with_numerical(line)]
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
67
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
68
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
69 # 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
70 g = SeqIO.to_dict(SeqIO.parse(genomefasta, "fasta"))
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
71 genome = defaultdict(dict)
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
72
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
73 for chr in g.keys():
15
2e3d976e7d5d planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 03183e29f807ec33548016a7c4144f52720b7b9e
artbio
parents: 14
diff changeset
74 genome[chr]['sequence'] = str(g[chr].seq)
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
75 genome[chr]['length'] = len(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
76
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
77 # Build a bedfile of repeatcoordinates to use by RepEnrich region_sorter
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
78 repeat_elements = set()
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
79 rep_coords = defaultdict(list) # Merged dictionary for coordinates
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
80
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
81 with open('repnames.bed', 'w') as fout:
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
82 f_in = import_text(annotation_file, ' ')
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
83 for line in f_in:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
84 repname = line[9].translate(str.maketrans('()/', '___'))
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
85 repeat_elements.add(repname)
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
86 repchr, repstart, repend = line[4], line[5], line[6]
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
87 fout.write(f"{repchr}\t{repstart}\t{repend}\t{repname}\n")
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
88 rep_coords[repname].extend([repchr, repstart, repend])
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
89 # repeat_elements now contains the unique repeat names
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
90 # rep_coords is a dictionary where keys are repeat names and values are lists
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
91 # containing chromosome, start, and end coordinates for each repeat instance
0
f6f0f1e5e940 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/repenrich commit 61e203df0be5ed877ff92b917c7cde6eeeab8310
artbio
parents:
diff changeset
92
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
93 # sort repeat_elements and print them in repeatIDs.txt
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
94 with open('repeatIDs.txt', 'w') as fout:
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
95 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
96 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
97
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
98 # 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
99 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
100
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
101 # generate metagenomes and save them to FASTA files for bowtie build
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
102 for repname in rep_coords:
15
2e3d976e7d5d planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 03183e29f807ec33548016a7c4144f52720b7b9e
artbio
parents: 14
diff changeset
103 genomes_list = []
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
104 # iterating coordinate list by block of 3 (chr, start, end)
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
105 block = 3
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
106 for i in range(0, len(rep_coords[repname]) - block + 1, block):
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
107 batch = rep_coords[repname][i:i+block]
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
108 chromosome = batch[0]
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
109 start = max(int(batch[1]) - flankingl, 0)
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
110 end = min(int(batch[2]) + flankingl,
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
111 int(genome[chromosome]['length'])-1) + 1
15
2e3d976e7d5d planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 03183e29f807ec33548016a7c4144f52720b7b9e
artbio
parents: 14
diff changeset
112 genomes_list.append(genome[chromosome]['sequence'][start:end])
2e3d976e7d5d planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 03183e29f807ec33548016a7c4144f52720b7b9e
artbio
parents: 14
diff changeset
113 metagenome = spacer.join(genomes_list)
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
114 # Create Fasta of repeat pseudogenome
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
115 fastafilename = f"{repname}.fa"
12
89e05f831259 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 212b838f614f1f7b8e770473c026d9c1180722df
artbio
parents: 11
diff changeset
116 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
117 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
118
13
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
119
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
120 def bowtie_build(args):
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
121 """
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
122 Function to be executed in parallel by ProcessPoolExecutor.
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
123 """
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
124 try:
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
125 bowtie_base, fasta = args
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
126 command = shlex.split(f"bowtie-build -f {fasta} {bowtie_base}")
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
127 squash = subprocess.run(command, capture_output=True, text=True)
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
128 return squash.stdout
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
129 except Exception as e:
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
130 return str(e)
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
131
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
132
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
133 args_list = [(name, f"{name}.fa") for name in rep_coords]
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
134 with ProcessPoolExecutor(max_workers=cpus) as executor:
530626b0757c planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit df6b9491ad06e8a85e67c663b68db3cce3eb0115
artbio
parents: 12
diff changeset
135 executor.map(bowtie_build, args_list)