Mercurial > repos > brinkmanlab > make_unique_id
comparison make_unique_id.py @ 3:a2258ce2d58c draft
"planemo upload for repository https://github.com/brinkmanlab/galaxy-tools/tree/master/make_unique_id commit 08b7b36e1eba516e4d7eb97086baf7ec8d177c5d"
author | brinkmanlab |
---|---|
date | Mon, 15 Jun 2020 19:32:32 -0400 |
parents | c8bda09480ae |
children | f2656e644641 |
comparison
equal
deleted
inserted
replaced
2:c8bda09480ae | 3:a2258ce2d58c |
---|---|
27 | 27 |
28 format = sys.argv[1] | 28 format = sys.argv[1] |
29 ids = defaultdict(int) | 29 ids = defaultdict(int) |
30 | 30 |
31 def makeUnique(seq): | 31 def makeUnique(seq): |
32 newid = seq.id[:16] # Genbank has a max length for the id and sequence length number, truncate the sequence id if too long | 32 seqlenlen = len(str(len(seq))) |
33 newid = seq.id[:26 - seqlenlen] # Genbank has a max length for the id and sequence length number, truncate the sequence id if too long | |
33 count = ids[newid] | 34 count = ids[newid] |
34 ids[newid] += 1 | 35 ids[newid] += 1 |
35 if count: | 36 if count: |
36 suffix = "_" + str(count) | 37 suffix = "_" + str(count) |
37 seqlenlen = len(str(len(seq))) | 38 if len(newid) + len(suffix) + 1 + seqlenlen > 26: |
38 if len(newid) + len(suffix) + 1 + seqlenlen > 16: | 39 newid = newid[:25 - seqlenlen - len(suffix)] |
39 newid = newid[:16 - seqlenlen - len(suffix)] | |
40 | 40 |
41 newid += suffix | 41 newid += suffix |
42 seq.name += suffix | 42 seq.name += suffix |
43 | 43 |
44 if seq.id != newid: | 44 if seq.id != newid: |
45 print(f"{seq.id}\t{newid}") | 45 print(f"{seq.id}\t{newid}") |
46 | 46 |
47 seq.id = newid | 47 seq.id = newid |
48 return seq | 48 return seq |
49 | 49 |
50 | |
50 paths = iter(sys.argv[2:]) | 51 paths = iter(sys.argv[2:]) |
51 | 52 |
52 for input, output in zip(paths, paths): | 53 for input, output in zip(paths, paths): |
53 SeqIO.write( | 54 SeqIO.write( |
54 map(makeUnique, SeqIO.parse(input, format)), | 55 map(makeUnique, SeqIO.parse(input, format)), |
55 output, | 56 output, |
56 format | 57 format |
57 ) | 58 ) |
58 | 59 |
60 |