Mercurial > repos > vipints > rdiff

comparison 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
comparison
equal deleted inserted replaced
0:0f80a5141704 2:233c30f91d66
1 #!/usr/bin/env python
2 """
3 Common utility functions
4 """
5
6 import os
7 import re
8 import sys
9 import gzip
10 import bz2
11 import numpy
12
13 def init_gene_DE():
14 """
15 Initializing the gene structure for DE
16 """
17 gene_det = [('id', 'f8'),
18 ('chr', 'S15'),
19 ('exons', numpy.dtype),
20 ('gene_info', numpy.dtype),
21 ('is_alt_spliced', 'f8'),
22 ('name', 'S25'),
23 ('source', 'S25'),
24 ('start', 'f8'),
25 ('stop', 'f8'),
26 ('strand', 'S2'),
27 ('transcripts', numpy.dtype)]
28
29 return gene_det
30
31 def _open_file(fname):
32 """
33 Open the file (supports .gz .bz2) and returns the handler
34 """
35 try:
36 if os.path.splitext(fname)[1] == ".gz":
37 FH = gzip.open(fname, 'rb')
38 elif os.path.splitext(fname)[1] == ".bz2":
39 FH = bz2.BZ2File(fname, 'rb')
40 else:
41 FH = open(fname, 'rU')
42 except Exception as error:
43 sys.exit(error)
44 return FH
45
46 def make_Exon_cod(strand_p, five_p_utr, cds_cod, three_p_utr):
47 """
48 Create exon cordinates from UTR's and CDS region
49 """
50 exon_pos = []
51 if strand_p == '+':
52 utr5_start, utr5_end = 0, 0
53 if five_p_utr != []:
54 utr5_start, utr5_end = five_p_utr[-1][0], five_p_utr[-1][1]
55 cds_5start, cds_5end = cds_cod[0][0], cds_cod[0][1]
56 jun_exon = []
57 if cds_5start-utr5_end == 0 or cds_5start-utr5_end == 1:
58 jun_exon = [utr5_start, cds_5end]
59 if len(cds_cod) == 1:
60 five_prime_flag = 0
61 if jun_exon != []:
62 five_p_utr = five_p_utr[:-1]
63 five_prime_flag = 1
64 for utr5 in five_p_utr:
65 exon_pos.append(utr5)
66 jun_exon = []
67 utr3_start, utr3_end = 0, 0
68 if three_p_utr != []:
69 utr3_start = three_p_utr[0][0]
70 utr3_end = three_p_utr[0][1]
71 if utr3_start-cds_5end == 0 or utr3_start-cds_5end == 1:
72 jun_exon = [cds_5start, utr3_end]
73 three_prime_flag = 0
74 if jun_exon != []:
75 cds_cod = cds_cod[:-1]
76 three_p_utr = three_p_utr[1:]
77 three_prime_flag = 1
78 if five_prime_flag == 1 and three_prime_flag == 1:
79 exon_pos.append([utr5_start, utr3_end])
80 if five_prime_flag == 1 and three_prime_flag == 0:
81 exon_pos.append([utr5_start, cds_5end])
82 cds_cod = cds_cod[:-1]
83 if five_prime_flag == 0 and three_prime_flag == 1:
84 exon_pos.append([cds_5start, utr3_end])
85 for cds in cds_cod:
86 exon_pos.append(cds)
87 for utr3 in three_p_utr:
88 exon_pos.append(utr3)
89 else:
90 if jun_exon != []:
91 five_p_utr = five_p_utr[:-1]
92 cds_cod = cds_cod[1:]
93 for utr5 in five_p_utr:
94 exon_pos.append(utr5)
95 exon_pos.append(jun_exon) if jun_exon != [] else ''
96 jun_exon = []
97 utr3_start, utr3_end = 0, 0
98 if three_p_utr != []:
99 utr3_start = three_p_utr[0][0]
100 utr3_end = three_p_utr[0][1]
101 cds_3start = cds_cod[-1][0]
102 cds_3end = cds_cod[-1][1]
103 if utr3_start-cds_3end == 0 or utr3_start-cds_3end == 1:
104 jun_exon = [cds_3start, utr3_end]
105 if jun_exon != []:
106 cds_cod = cds_cod[:-1]
107 three_p_utr = three_p_utr[1:]
108 for cds in cds_cod:
109 exon_pos.append(cds)
110 exon_pos.append(jun_exon) if jun_exon != [] else ''
111 for utr3 in three_p_utr:
112 exon_pos.append(utr3)
113 elif strand_p == '-':
114 utr3_start, utr3_end = 0, 0
115 if three_p_utr != []:
116 utr3_start = three_p_utr[-1][0]
117 utr3_end = three_p_utr[-1][1]
118 cds_3start = cds_cod[0][0]
119 cds_3end = cds_cod[0][1]
120 jun_exon = []
121 if cds_3start-utr3_end == 0 or cds_3start-utr3_end == 1:
122 jun_exon = [utr3_start, cds_3end]
123 if len(cds_cod) == 1:
124 three_prime_flag = 0
125 if jun_exon != []:
126 three_p_utr = three_p_utr[:-1]
127 three_prime_flag = 1
128 for utr3 in three_p_utr:
129 exon_pos.append(utr3)
130 jun_exon = []
131 (utr5_start, utr5_end) = (0, 0)
132 if five_p_utr != []:
133 utr5_start = five_p_utr[0][0]
134 utr5_end = five_p_utr[0][1]
135 if utr5_start-cds_3end == 0 or utr5_start-cds_3end == 1:
136 jun_exon = [cds_3start, utr5_end]
137 five_prime_flag = 0
138 if jun_exon != []:
139 cds_cod = cds_cod[:-1]
140 five_p_utr = five_p_utr[1:]
141 five_prime_flag = 1
142 if three_prime_flag == 1 and five_prime_flag == 1:
143 exon_pos.append([utr3_start, utr5_end])
144 if three_prime_flag == 1 and five_prime_flag == 0:
145 exon_pos.append([utr3_start, cds_3end])
146 cds_cod = cds_cod[:-1]
147 if three_prime_flag == 0 and five_prime_flag == 1:
148 exon_pos.append([cds_3start, utr5_end])
149 for cds in cds_cod:
150 exon_pos.append(cds)
151 for utr5 in five_p_utr:
152 exon_pos.append(utr5)
153 else:
154 if jun_exon != []:
155 three_p_utr = three_p_utr[:-1]
156 cds_cod = cds_cod[1:]
157 for utr3 in three_p_utr:
158 exon_pos.append(utr3)
159 if jun_exon != []:
160 exon_pos.append(jun_exon)
161 jun_exon = []
162 (utr5_start, utr5_end) = (0, 0)
163 if five_p_utr != []:
164 utr5_start = five_p_utr[0][0]
165 utr5_end = five_p_utr[0][1]
166 cds_5start = cds_cod[-1][0]
167 cds_5end = cds_cod[-1][1]
168 if utr5_start-cds_5end == 0 or utr5_start-cds_5end == 1:
169 jun_exon = [cds_5start, utr5_end]
170 if jun_exon != []:
171 cds_cod = cds_cod[:-1]
172 five_p_utr = five_p_utr[1:]
173 for cds in cds_cod:
174 exon_pos.append(cds)
175 if jun_exon != []:
176 exon_pos.append(jun_exon)
177 for utr5 in five_p_utr:
178 exon_pos.append(utr5)
179 return exon_pos