Mercurial > repos > vipints > fml_gff3togtf

annotate GFFtools-GX/gbk_to_gff.py @ 3:ff2c2e6f4ab3

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded version 2.0.0 of gfftools ready to import to local instance
author vipints
date Wed, 11 Jun 2014 16:29:25 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
1 #!/usr/bin/env python
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
2 """
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
3 Convert data from Genbank format to GFF.
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
4
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
5 Usage:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
6 python gbk_to_gff.py in.gbk > out.gff
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
7
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
8 Requirements:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
9 BioPython:- http://biopython.org/
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
10 helper.py : https://github.com/vipints/GFFtools-GX/blob/master/helper.py
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
11
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
12 Copyright (C)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
13 2009-2012 Friedrich Miescher Laboratory of the Max Planck Society, Tubingen, Germany.
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
14 2012-2014 Memorial Sloan Kettering Cancer Center New York City, USA.
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
15 """
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
16
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
17 import os
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
18 import re
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
19 import sys
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
20 import collections
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
21 from Bio import SeqIO
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
22 import helper
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
23
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
24 def feature_table(chr_id, source, orient, genes, transcripts, cds, exons, unk):
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
25 """
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
26 Write the feature information
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
27 """
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
28
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
29 for gname, ginfo in genes.items():
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
30 line = [str(chr_id),
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
31 'gbk_to_gff',
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
32 ginfo[3],
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
33 str(ginfo[0]),
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
34 str(ginfo[1]),
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
35 '.',
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
36 ginfo[2],
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
37 '.',
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
38 'ID=%s;Name=%s' % (str(gname), str(gname))]
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
39 print '\t'.join(line)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
40 ## construct the transcript line is not defined in the original file
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
41 t_line = [str(chr_id), 'gbk_to_gff', source, 0, 1, '.', ginfo[2], '.']
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
42
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
43 if not transcripts:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
44 t_line.append('ID=Transcript:%s;Parent=%s' % (str(gname), str(gname)))
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
45
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
46 if exons: ## get the entire transcript region from the defined feature
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
47 t_line[3] = str(exons[gname][0][0])
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
48 t_line[4] = str(exons[gname][0][-1])
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
49 elif cds:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
50 t_line[3] = str(cds[gname][0][0])
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
51 t_line[4] = str(cds[gname][0][-1])
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
52 print '\t'.join(t_line)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
53
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
54 if exons:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
55 exon_line_print(t_line, exons[gname], 'Transcript:'+str(gname), 'exon')
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
56
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
57 if cds:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
58 exon_line_print(t_line, cds[gname], 'Transcript:'+str(gname), 'CDS')
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
59 if not exons:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
60 exon_line_print(t_line, cds[gname], 'Transcript:'+str(gname), 'exon')
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
61
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
62 else: ## transcript is defined
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
63 for idx in transcripts[gname]:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
64 t_line[2] = idx[3]
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
65 t_line[3] = str(idx[0])
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
66 t_line[4] = str(idx[1])
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
67 t_line.append('ID='+str(idx[2])+';Parent='+str(gname))
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
68 print '\t'.join(t_line)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
69
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
70 ## feature line print call
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
71 if exons:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
72 exon_line_print(t_line, exons[gname], str(idx[2]), 'exon')
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
73 if cds:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
74 exon_line_print(t_line, cds[gname], str(idx[2]), 'CDS')
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
75 if not exons:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
76 exon_line_print(t_line, cds[gname], str(idx[2]), 'exon')
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
77
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
78 if len(genes) == 0: ## feature entry with fragment information
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
79
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
80 line = [str(chr_id), 'gbk_to_gff', source, 0, 1, '.', orient, '.']
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
81 fStart = fStop = None
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
82
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
83 for eid, ex in cds.items():
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
84 fStart = ex[0][0]
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
85 fStop = ex[0][-1]
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
86
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
87 for eid, ex in exons.items():
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
88 fStart = ex[0][0]
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
89 fStop = ex[0][-1]
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
90
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
91 if fStart or fStart:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
92
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
93 line[2] = 'gene'
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
94 line[3] = str(fStart)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
95 line[4] = str(fStop)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
96 line.append('ID=Unknown_Gene_' + str(unk) + ';Name=Unknown_Gene_' + str(unk))
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
97 print "\t".join(line)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
98
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
99 if not cds:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
100 line[2] = 'transcript'
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
101 else:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
102 line[2] = 'mRNA'
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
103
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
104 line[8] = 'ID=Unknown_Transcript_' + str(unk) + ';Parent=Unknown_Gene_' + str(unk)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
105 print "\t".join(line)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
106
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
107 if exons:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
108 exon_line_print(line, cds[None], 'Unknown_Transcript_' + str(unk), 'exon')
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
109
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
110 if cds:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
111 exon_line_print(line, cds[None], 'Unknown_Transcript_' + str(unk), 'CDS')
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
112 if not exons:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
113 exon_line_print(line, cds[None], 'Unknown_Transcript_' + str(unk), 'exon')
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
114
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
115 unk +=1
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
116
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
117 return unk
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
118
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
119 def exon_line_print(temp_line, trx_exons, parent, ftype):
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
120 """
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
121 Print the EXON feature line
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
122 """
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
123
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
124 for ex in trx_exons:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
125 temp_line[2] = ftype
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
126 temp_line[3] = str(ex[0])
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
127 temp_line[4] = str(ex[1])
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
128 temp_line[8] = 'Parent=%s' % parent
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
129 print '\t'.join(temp_line)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
130
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
131 def gbk_parse(fname):
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
132 """
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
133 Extract genome annotation recods from genbank format
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
134
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
135 @args fname: gbk file name
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
136 @type fname: str
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
137 """
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
138
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
139 fhand = helper.open_file(gbkfname)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
140 unk = 1
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
141
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
142 for record in SeqIO.parse(fhand, "genbank"):
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
143
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
144 gene_tags = dict()
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
145 tx_tags = collections.defaultdict(list)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
146 exon = collections.defaultdict(list)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
147 cds = collections.defaultdict(list)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
148 mol_type, chr_id = None, None
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
149
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
150 for rec in record.features:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
151
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
152 if rec.type == 'source':
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
153 try:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
154 mol_type = rec.qualifiers['mol_type'][0]
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
155 except:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
156 mol_type = '.'
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
157 pass
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
158 try:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
159 chr_id = rec.qualifiers['chromosome'][0]
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
160 except:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
161 chr_id = record.name
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
162 continue
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
163
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
164 strand='-'
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
165 strand='+' if rec.strand>0 else strand
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
166
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
167 fid = None
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
168 try:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
169 fid = rec.qualifiers['gene'][0]
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
170 except:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
171 pass
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
172
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
173 transcript_id = None
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
174 try:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
175 transcript_id = rec.qualifiers['transcript_id'][0]
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
176 except:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
177 pass
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
178
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
179 if re.search(r'gene', rec.type):
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
180 gene_tags[fid] = (rec.location._start.position+1,
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
181 rec.location._end.position,
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
182 strand,
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
183 rec.type
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
184 )
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
185 elif rec.type == 'exon':
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
186 exon[fid].append((rec.location._start.position+1,
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
187 rec.location._end.position))
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
188 elif rec.type=='CDS':
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
189 cds[fid].append((rec.location._start.position+1,
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
190 rec.location._end.position))
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
191 else:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
192 # get all transcripts
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
193 if transcript_id:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
194 tx_tags[fid].append((rec.location._start.position+1,
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
195 rec.location._end.position,
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
196 transcript_id,
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
197 rec.type))
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
198 # record extracted, generate feature table
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
199 unk = feature_table(chr_id, mol_type, strand, gene_tags, tx_tags, cds, exon, unk)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
200
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
201 fhand.close()
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
202
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
203
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
204 if __name__=='__main__':
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
205
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
206 try:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
207 gbkfname = sys.argv[1]
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
208 except:
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
209 print __doc__
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
210 sys.exit(-1)
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
211
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
212 ## extract gbk records
ff2c2e6f4ab3 Uploaded version 2.0.0 of gfftools ready to import to local instance
vipints
parents:
diff changeset
213 gbk_parse(gbkfname)