annotate glimmer2seq.py @ 0:12a24734bba8 draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
author iuc
date Tue, 28 Nov 2017 09:56:20 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
1 #!/usr/bin/env python
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
2 """
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
3 Input: DNA FASTA file + Glimmer ORF file
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
4 Output: ORF sequences as FASTA file
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
5 Author: Bjoern Gruening
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
6 """
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
7 import sys
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
8
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
9 from Bio import SeqIO
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
10 from Bio.SeqRecord import SeqRecord
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
11
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
12
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
13 def glimmer2seq(glimmer_prediction=sys.argv[1], genome_sequence=sys.argv[2], outfile=sys.argv[3]):
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
14 if len(sys.argv) >= 4:
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
15 glimmerfile = open(glimmer_prediction, "r")
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
16 sequence = open(genome_sequence)
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
17 else:
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
18 print("Missing input values.")
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
19 sys.exit()
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
20
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
21 fastafile = SeqIO.parse(sequence, "fasta")
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
22
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
23 sequences = dict()
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
24 seq_records = list()
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
25 for entry in fastafile:
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
26 sequences[entry.description] = entry
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
27
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
28 for line in glimmerfile:
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
29 if line.startswith('>'):
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
30 entry = sequences[line[1:].strip()]
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
31 else:
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
32 orf_start = int(line[8:17])
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
33 orf_end = int(line[18:26])
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
34
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
35 orf_name = line[0:8]
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
36 if orf_start <= orf_end:
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
37 seq_records.append(SeqRecord(entry.seq[orf_start - 1:orf_end], id=orf_name, description=entry.description))
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
38 else:
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
39 seq_records.append(SeqRecord(entry.seq[orf_end - 1:orf_start].reverse_complement(), id=orf_name, description=entry.description))
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
40
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
41 SeqIO.write(seq_records, outfile, "fasta")
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
42 glimmerfile.close()
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
43 sequence.close()
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
44
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
45
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
46 if __name__ == "__main__":
12a24734bba8 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/glimmer commit 37388949e348d221170659bbee547bf4ac67ef1a
iuc
parents:
diff changeset
47 glimmer2seq()