Mercurial > repos > cpt > cpt_gbk_to_5col

annotate gbk_to_five_col.py @ 2:8f5f2142fc2b draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc
author cpt
date Fri, 05 Jan 2024 05:51:38 +0000
parents 1bdd481d5c25
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
1 #!/usr/bin/env python
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
2 import BIO_FIX_TOPO # NOQA
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
3 import argparse
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
4 import logging
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
5 from Bio import SeqIO
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
6
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
7 logging.basicConfig(level=logging.INFO)
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
8 log = logging.getLogger()
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
9
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
10
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
11 # Read in Genbank file and parse features
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
12 # Output features into Five Column format
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
13
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
14 """
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
15 >Feature SeqID
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
16 Line 1
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
17 Column 1: Start location (first nucleotide) of a feature
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
18 Column 2: Stop location (last nucleotide) of a feature
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
19 Column 3: Feature name (for example, 'CDS' or 'mRNA' or 'rRNA' or 'gene' or 'exon')
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
20 Line2:
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
21 Column 4: Qualifier name (for example, 'product' or 'number' or 'gene' or 'note')
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
22 Column 5: Qualifier value
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
23
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
24 Repeat for each feature in a seq
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
25 Repeat Line 2 for each qualifier in a feature
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
26 """
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
27
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
28
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
29 def gbk_to_5col(genbank):
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
30 """Converts genbank to BankIt five column format"""
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
31 for record in SeqIO.parse(genbank, "genbank"):
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
32 print(">Feature %s" % record.id)
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
33 for feature in record.features:
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
34 if feature.type == "source":
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
35 continue
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
36 else:
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
37 for index, part in enumerate(feature.location.parts):
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
38 if part.strand > 0:
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
39 start = int(part.start) + 1
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
40 end = int(part.end)
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
41 else:
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
42 start = int(part.end)
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
43 end = int(part.start) + 1
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
44 if index == 0:
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
45 name = feature.type
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
46 print("%d\t%d\t%s" % (start, end, name))
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
47 else:
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
48 print("%d\t%d" % (start, end))
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
49 for (qualifier, values) in feature.qualifiers.items():
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
50 for value in values:
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
51 print("\t\t\t%s\t%s" % (qualifier, value))
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
52
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
53
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
54 if __name__ == "__main__":
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
55 parser = argparse.ArgumentParser(
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
56 description="Convert a Genbank file into five column format"
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
57 )
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
58 parser.add_argument("genbank", type=argparse.FileType("r"), help="Genbank file")
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
59
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
60 args = vars(parser.parse_args())
1bdd481d5c25 planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff changeset
61 gbk_to_5col(**args)