annotate cut_up_fasta.py @ 2:92a2d9109528 draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 77dea70f156d56c83b6851b61dc997d2b344bdc9
author iuc
date Fri, 01 Jul 2022 14:16:40 +0000
parents a04028a8181d
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
1 #!/usr/bin/env python
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
2
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
3 import argparse
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
4 import gzip
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
5 from functools import partial
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
6
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
7 from Bio import SeqIO
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
8
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
9
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
10 def cut_up_fasta(input_fasta, chunk_size, overlap, merge_last, output_fasta, output_bed, gzipped):
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
11 if gzipped:
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
12 _open = partial(gzip.open, mode='rt')
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
13 else:
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
14 _open = open
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
15
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
16 fasta_fh = open(output_fasta, 'w')
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
17
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
18 if output_bed is not None:
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
19 bed_fh = open(output_bed, 'w')
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
20
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
21 with _open(input_fasta) as input_fh:
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
22 for record in SeqIO.parse(input_fh, "fasta"):
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
23 if (not merge_last and len(record.seq) > chunk_size) or (merge_last and len(record.seq) >= 2 * chunk_size):
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
24 for index, split_seq in enumerate(chunks(record.seq, chunk_size, overlap, merge_last)):
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
25 fasta_fh.write(f">{record.id}.concoct_part_{index}\n{split_seq}\n")
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
26 if output_bed is not None:
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
27 bed_fh.write(f"{record.id}\t{chunk_size * index}\t{chunk_size * index + len(split_seq)}\t{record.id}.concoct_part_{index}\n")
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
28 else:
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
29 fasta_fh.write(f">{record.id}.concoct_part_0\n{record.seq}\n")
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
30 if output_bed is not None:
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
31 bed_fh.write(f"{record.id}\t0\t{len(record.seq)}\t{record.id}.concoct_part_0\n")
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
32 if output_bed is not None:
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
33 bed_fh.close()
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
34
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
35
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
36 def chunks(seq, chunk_size, overlap_size, merge_last):
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
37 # Yield successive chunk_size-sized chunks from seq
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
38 # with given overlap overlap_size between the chunks.
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
39 assert chunk_size > overlap_size
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
40 if merge_last:
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
41 for i in range(0, len(seq) - chunk_size + 1, chunk_size - overlap_size):
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
42 yield seq[i:i + chunk_size] if i + chunk_size + chunk_size - overlap_size <= len(seq) else seq[i:]
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
43 else:
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
44 for i in range(0, len(seq), chunk_size - overlap_size):
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
45 yield seq[i:i + chunk_size]
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
46
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
47
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
48 parser = argparse.ArgumentParser()
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
49 parser.add_argument("--input_fasta", action="store", dest="input_fasta", help="Fasta files with contigs")
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
50 parser.add_argument("--gzipped", action="store_true", dest="gzipped", help="Input file is gzipped")
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
51 parser.add_argument("--chunk_size", action="store", dest="chunk_size", type=int, help="Chunk size\n")
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
52 parser.add_argument("--overlap_size", action="store", dest="overlap_size", type=int, help="Overlap size\n")
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
53 parser.add_argument("--merge_last", default=False, action="store_true", dest="merge_last", help="Concatenate final part to last contig\n")
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
54 parser.add_argument("--output_bed", action="store", dest="output_bed", default=None, help="BED file to be created with exact regions of the original contigs corresponding to the newly created contigs")
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
55 parser.add_argument("--output_fasta", action="store", dest="output_fasta", help="Output fasta file with cut contigs")
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
56
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
57 args = parser.parse_args()
a04028a8181d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/concoct commit 40a09cbfd6052f7b0295946621db1bdf58228b09"
iuc
parents:
diff changeset
58 cut_up_fasta(args.input_fasta, args.chunk_size, args.overlap_size, args.merge_last, args.output_fasta, args.output_bed, args.gzipped)