Mercurial > repos > cpt > cpt_blasttab_dice_filter
annotate blasttab_dice_filter.py @ 6:f6d253d91757 draft default tip
planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc
author | cpt |
---|---|
date | Fri, 05 Jan 2024 05:49:45 +0000 |
parents | 99baf3ee2a2b |
children |
rev | line source |
---|---|
5
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
1 #!/usr/bin/env python |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
2 import argparse |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
3 import logging |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
4 |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
5 logging.basicConfig(level=logging.INFO) |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
6 log = logging.getLogger(name="blasttab2gff3") |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
7 |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
8 __doc__ = """ |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
9 Blast TSV files, when transformed to GFF3, do not normally show gaps in the |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
10 blast hits. This tool aims to fill that "gap". |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
11 """ |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
12 |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
13 |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
14 def blasttsv2gff3(blasttsv, min_dice=50): |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
15 # 01 Query Seq-id (ID of your sequence) |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
16 # 02 Subject Seq-id (ID of the database hit) |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
17 # 03 Percentage of identical matches |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
18 # 04 Alignment length |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
19 # 05 Number of mismatches |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
20 # 06 Number of gap openings |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
21 # 07 Start of alignment in query |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
22 # 08 End of alignment in query |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
23 # 09 Start of alignment in subject (database hit) |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
24 # 10 End of alignment in subject (database hit) |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
25 # 11 Expectation value (E-value) |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
26 # 12 Bit score |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
27 # 13 All subject Seq-id(s), separated by a ';' |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
28 # 14 Raw score |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
29 # 15 Number of identical matches |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
30 # 16 Number of positive-scoring matches |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
31 # 17 Total number of gaps |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
32 # 18 Percentage of positive-scoring matches |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
33 # 19 Query frame |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
34 # 20 Subject frame |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
35 # 21 Aligned part of query sequence |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
36 # 22 Aligned part of subject sequence |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
37 # 23 Query sequence length |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
38 # 24 Subject sequence length |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
39 # 25 All subject title(s), separated by a '<>' |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
40 |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
41 for line in blasttsv: |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
42 line = line.strip("\n") |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
43 data = line.split("\t") |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
44 dice = 2 * float(data[14]) / (float(data[22]) + float(data[23])) |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
45 |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
46 if dice >= min_dice: |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
47 yield line |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
48 |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
49 |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
50 if __name__ == "__main__": |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
51 parser = argparse.ArgumentParser(description="Convert Blast TSV to gapped GFF3") |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
52 parser.add_argument( |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
53 "blasttsv", type=argparse.FileType("r"), help="Blast TSV Output" |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
54 ) |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
55 parser.add_argument( |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
56 "--min_dice", type=float, help="Minimum dice score", default=0.5 |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
57 ) |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
58 args = parser.parse_args() |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
59 |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
60 for line in blasttsv2gff3(**vars(args)): |
99baf3ee2a2b
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
cpt
parents:
diff
changeset
|
61 print(line) |