Mercurial > repos > vipints > rdiff

annotate rDiff/tools/helper.py @ 2:233c30f91d66

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
author vipints <vipin@cbio.mskcc.org>
date Tue, 08 Oct 2013 07:15:44 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
1 #!/usr/bin/env python
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
2 """
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
3 Common utility functions
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
4 """
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
5
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
6 import os
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
7 import re
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
8 import sys
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
9 import gzip
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
10 import bz2
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
11 import numpy
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
12
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
13 def init_gene_DE():
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
14 """
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
15 Initializing the gene structure for DE
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
16 """
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
17 gene_det = [('id', 'f8'),
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
18 ('chr', 'S15'),
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
19 ('exons', numpy.dtype),
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
20 ('gene_info', numpy.dtype),
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
21 ('is_alt_spliced', 'f8'),
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
22 ('name', 'S25'),
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
23 ('source', 'S25'),
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
24 ('start', 'f8'),
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
25 ('stop', 'f8'),
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
26 ('strand', 'S2'),
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
27 ('transcripts', numpy.dtype)]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
28
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
29 return gene_det
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
30
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
31 def _open_file(fname):
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
32 """
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
33 Open the file (supports .gz .bz2) and returns the handler
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
34 """
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
35 try:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
36 if os.path.splitext(fname)[1] == ".gz":
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
37 FH = gzip.open(fname, 'rb')
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
38 elif os.path.splitext(fname)[1] == ".bz2":
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
39 FH = bz2.BZ2File(fname, 'rb')
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
40 else:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
41 FH = open(fname, 'rU')
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
42 except Exception as error:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
43 sys.exit(error)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
44 return FH
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
45
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
46 def make_Exon_cod(strand_p, five_p_utr, cds_cod, three_p_utr):
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
47 """
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
48 Create exon cordinates from UTR's and CDS region
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
49 """
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
50 exon_pos = []
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
51 if strand_p == '+':
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
52 utr5_start, utr5_end = 0, 0
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
53 if five_p_utr != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
54 utr5_start, utr5_end = five_p_utr[-1][0], five_p_utr[-1][1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
55 cds_5start, cds_5end = cds_cod[0][0], cds_cod[0][1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
56 jun_exon = []
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
57 if cds_5start-utr5_end == 0 or cds_5start-utr5_end == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
58 jun_exon = [utr5_start, cds_5end]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
59 if len(cds_cod) == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
60 five_prime_flag = 0
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
61 if jun_exon != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
62 five_p_utr = five_p_utr[:-1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
63 five_prime_flag = 1
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
64 for utr5 in five_p_utr:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
65 exon_pos.append(utr5)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
66 jun_exon = []
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
67 utr3_start, utr3_end = 0, 0
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
68 if three_p_utr != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
69 utr3_start = three_p_utr[0][0]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
70 utr3_end = three_p_utr[0][1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
71 if utr3_start-cds_5end == 0 or utr3_start-cds_5end == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
72 jun_exon = [cds_5start, utr3_end]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
73 three_prime_flag = 0
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
74 if jun_exon != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
75 cds_cod = cds_cod[:-1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
76 three_p_utr = three_p_utr[1:]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
77 three_prime_flag = 1
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
78 if five_prime_flag == 1 and three_prime_flag == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
79 exon_pos.append([utr5_start, utr3_end])
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
80 if five_prime_flag == 1 and three_prime_flag == 0:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
81 exon_pos.append([utr5_start, cds_5end])
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
82 cds_cod = cds_cod[:-1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
83 if five_prime_flag == 0 and three_prime_flag == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
84 exon_pos.append([cds_5start, utr3_end])
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
85 for cds in cds_cod:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
86 exon_pos.append(cds)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
87 for utr3 in three_p_utr:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
88 exon_pos.append(utr3)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
89 else:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
90 if jun_exon != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
91 five_p_utr = five_p_utr[:-1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
92 cds_cod = cds_cod[1:]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
93 for utr5 in five_p_utr:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
94 exon_pos.append(utr5)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
95 exon_pos.append(jun_exon) if jun_exon != [] else ''
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
96 jun_exon = []
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
97 utr3_start, utr3_end = 0, 0
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
98 if three_p_utr != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
99 utr3_start = three_p_utr[0][0]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
100 utr3_end = three_p_utr[0][1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
101 cds_3start = cds_cod[-1][0]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
102 cds_3end = cds_cod[-1][1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
103 if utr3_start-cds_3end == 0 or utr3_start-cds_3end == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
104 jun_exon = [cds_3start, utr3_end]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
105 if jun_exon != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
106 cds_cod = cds_cod[:-1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
107 three_p_utr = three_p_utr[1:]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
108 for cds in cds_cod:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
109 exon_pos.append(cds)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
110 exon_pos.append(jun_exon) if jun_exon != [] else ''
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
111 for utr3 in three_p_utr:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
112 exon_pos.append(utr3)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
113 elif strand_p == '-':
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
114 utr3_start, utr3_end = 0, 0
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
115 if three_p_utr != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
116 utr3_start = three_p_utr[-1][0]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
117 utr3_end = three_p_utr[-1][1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
118 cds_3start = cds_cod[0][0]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
119 cds_3end = cds_cod[0][1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
120 jun_exon = []
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
121 if cds_3start-utr3_end == 0 or cds_3start-utr3_end == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
122 jun_exon = [utr3_start, cds_3end]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
123 if len(cds_cod) == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
124 three_prime_flag = 0
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
125 if jun_exon != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
126 three_p_utr = three_p_utr[:-1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
127 three_prime_flag = 1
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
128 for utr3 in three_p_utr:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
129 exon_pos.append(utr3)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
130 jun_exon = []
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
131 (utr5_start, utr5_end) = (0, 0)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
132 if five_p_utr != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
133 utr5_start = five_p_utr[0][0]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
134 utr5_end = five_p_utr[0][1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
135 if utr5_start-cds_3end == 0 or utr5_start-cds_3end == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
136 jun_exon = [cds_3start, utr5_end]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
137 five_prime_flag = 0
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
138 if jun_exon != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
139 cds_cod = cds_cod[:-1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
140 five_p_utr = five_p_utr[1:]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
141 five_prime_flag = 1
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
142 if three_prime_flag == 1 and five_prime_flag == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
143 exon_pos.append([utr3_start, utr5_end])
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
144 if three_prime_flag == 1 and five_prime_flag == 0:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
145 exon_pos.append([utr3_start, cds_3end])
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
146 cds_cod = cds_cod[:-1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
147 if three_prime_flag == 0 and five_prime_flag == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
148 exon_pos.append([cds_3start, utr5_end])
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
149 for cds in cds_cod:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
150 exon_pos.append(cds)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
151 for utr5 in five_p_utr:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
152 exon_pos.append(utr5)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
153 else:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
154 if jun_exon != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
155 three_p_utr = three_p_utr[:-1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
156 cds_cod = cds_cod[1:]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
157 for utr3 in three_p_utr:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
158 exon_pos.append(utr3)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
159 if jun_exon != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
160 exon_pos.append(jun_exon)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
161 jun_exon = []
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
162 (utr5_start, utr5_end) = (0, 0)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
163 if five_p_utr != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
164 utr5_start = five_p_utr[0][0]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
165 utr5_end = five_p_utr[0][1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
166 cds_5start = cds_cod[-1][0]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
167 cds_5end = cds_cod[-1][1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
168 if utr5_start-cds_5end == 0 or utr5_start-cds_5end == 1:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
169 jun_exon = [cds_5start, utr5_end]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
170 if jun_exon != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
171 cds_cod = cds_cod[:-1]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
172 five_p_utr = five_p_utr[1:]
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
173 for cds in cds_cod:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
174 exon_pos.append(cds)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
175 if jun_exon != []:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
176 exon_pos.append(jun_exon)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
177 for utr5 in five_p_utr:
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
178 exon_pos.append(utr5)
233c30f91d66 updated python based GFF parsing module which will handle GTF/GFF/GFF3 file types
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
179 return exon_pos