annotate GFFParser.py @ 10:c42c69aa81f8

fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
author vipints <vipin@cbio.mskcc.org>
date Thu, 23 Apr 2015 18:01:45 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
10
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
1 #!/usr/bin/env python
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
2 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
3 Extract genome annotation from a GFF (a tab delimited format for storing sequence features and annotations) file.
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
4
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
5 Requirements:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
6 Numpy :- http://numpy.org/
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
7
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
8 Copyright (C)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
9
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
10 2009-2012 Friedrich Miescher Laboratory of the Max Planck Society, Tubingen, Germany.
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
11 2012-2015 Memorial Sloan Kettering Cancer Center, New York City, USA.
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
12 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
13
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
14 import re
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
15 import os
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
16 import sys
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
17 import urllib
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
18 import numpy as np
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
19 import helper as utils
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
20 from collections import defaultdict
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
21
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
22 def attribute_tags(col9):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
23 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
24 Split the key-value tags from the attribute column, it takes column number 9 from GTF/GFF file
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
25
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
26 @args col9: attribute column from GFF file
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
27 @type col9: str
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
28 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
29 info = defaultdict(list)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
30 is_gff = False
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
31
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
32 if not col9:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
33 return is_gff, info
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
34
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
35 # trim the line ending semi-colon ucsc may have some white-space
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
36 col9 = col9.rstrip(';| ')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
37 # attributes from 9th column
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
38 atbs = col9.split(" ; ")
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
39 if len(atbs) == 1:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
40 atbs = col9.split("; ")
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
41 if len(atbs) == 1:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
42 atbs = col9.split(";")
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
43 # check the GFF3 pattern which has key value pairs like:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
44 gff3_pat = re.compile("\w+=")
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
45 # sometime GTF have: gene_id uc002zkg.1;
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
46 gtf_pat = re.compile("\s?\w+\s")
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
47
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
48 key_vals = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
49
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
50 if gff3_pat.match(atbs[0]): # gff3 pattern
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
51 is_gff = True
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
52 key_vals = [at.split('=') for at in atbs]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
53 elif gtf_pat.match(atbs[0]): # gtf pattern
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
54 for at in atbs:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
55 key_vals.append(at.strip().split(" ",1))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
56 else:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
57 # to handle attribute column has only single value
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
58 key_vals.append(['ID', atbs[0]])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
59 # get key, val items
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
60 for item in key_vals:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
61 key, val = item
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
62 # replace the double qoutes from feature identifier
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
63 val = re.sub('"', '', val)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
64 # replace the web formating place holders to plain text format
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
65 info[key].extend([urllib.unquote(v) for v in val.split(',') if v])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
66
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
67 return is_gff, info
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
68
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
69 def spec_features_keywd(gff_parts):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
70 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
71 Specify the feature key word according to the GFF specifications
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
72
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
73 @args gff_parts: attribute field key
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
74 @type gff_parts: str
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
75 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
76 for t_id in ["transcript_id", "transcriptId", "proteinId"]:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
77 try:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
78 gff_parts["info"]["Parent"] = gff_parts["info"][t_id]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
79 break
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
80 except KeyError:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
81 pass
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
82 for g_id in ["gene_id", "geneid", "geneId", "name", "gene_name", "genename"]:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
83 try:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
84 gff_parts["info"]["GParent"] = gff_parts["info"][g_id]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
85 break
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
86 except KeyError:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
87 pass
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
88 ## TODO key words
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
89 for flat_name in ["Transcript", "CDS"]:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
90 if gff_parts["info"].has_key(flat_name):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
91 # parents
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
92 if gff_parts['type'] in [flat_name] or re.search(r'transcript', gff_parts['type'], re.IGNORECASE):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
93 if not gff_parts['id']:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
94 gff_parts['id'] = gff_parts['info'][flat_name][0]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
95 #gff_parts["info"]["ID"] = [gff_parts["id"]]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
96 # children
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
97 elif gff_parts["type"] in ["intron", "exon", "three_prime_UTR",
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
98 "coding_exon", "five_prime_UTR", "CDS", "stop_codon",
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
99 "start_codon"]:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
100 gff_parts["info"]["Parent"] = gff_parts["info"][flat_name]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
101 break
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
102 return gff_parts
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
103
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
104 def Parse(ga_file):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
105 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
106 Parsing GFF/GTF file based on feature relationship, it takes the input file.
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
107
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
108 @args ga_file: input file name
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
109 @type ga_file: str
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
110 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
111 child_map = defaultdict(list)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
112 parent_map = dict()
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
113
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
114 ga_handle = utils.open_file(ga_file)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
115
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
116 for rec in ga_handle:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
117 rec = rec.strip('\n\r')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
118
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
119 # skip empty line fasta identifier and commented line
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
120 if not rec or rec[0] in ['#', '>']:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
121 continue
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
122 # skip the genome sequence
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
123 if not re.search('\t', rec):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
124 continue
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
125
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
126 parts = rec.split('\t')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
127 assert len(parts) >= 8, rec
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
128
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
129 # process the attribute column (9th column)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
130 ftype, tags = attribute_tags(parts[-1])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
131 if not tags: # skip the line if no attribute column.
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
132 continue
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
133
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
134 # extract fields
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
135 if parts[1]:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
136 tags["source"] = parts[1]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
137 if parts[7]:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
138 tags["phase"] = parts[7]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
139
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
140 gff_info = dict()
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
141 gff_info['info'] = dict(tags)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
142 gff_info["is_gff3"] = ftype
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
143 gff_info['chr'] = parts[0]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
144 gff_info['score'] = parts[5]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
145
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
146 if parts[3] and parts[4]:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
147 gff_info['location'] = [int(parts[3]) ,
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
148 int(parts[4])]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
149 gff_info['type'] = parts[2]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
150 gff_info['id'] = tags.get('ID', [''])[0]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
151 if parts[6] in ['?', '.']:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
152 parts[6] = None
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
153 gff_info['strand'] = parts[6]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
154
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
155 # key word according to the GFF spec.
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
156 # is_gff3 flag is false check this condition and get the attribute fields
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
157 if not ftype:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
158 gff_info = spec_features_keywd(gff_info)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
159
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
160 # link the feature relationships
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
161 if gff_info['info'].has_key('Parent'):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
162 for p in gff_info['info']['Parent']:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
163 if p == gff_info['id']:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
164 gff_info['id'] = ''
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
165 break
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
166 rec_category = 'child'
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
167 elif gff_info['id']:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
168 rec_category = 'parent'
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
169 else:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
170 rec_category = 'record'
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
171
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
172 # depends on the record category organize the features
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
173 if rec_category == 'child':
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
174 for p in gff_info['info']['Parent']:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
175 # create the data structure based on source and feature id
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
176 child_map[(gff_info['chr'], gff_info['info']['source'], p)].append(
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
177 dict( type = gff_info['type'],
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
178 location = gff_info['location'],
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
179 strand = gff_info['strand'],
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
180 score = gff_info['score'],
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
181 ID = gff_info['id'],
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
182 gene_id = gff_info['info'].get('GParent', '')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
183 ))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
184 elif rec_category == 'parent':
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
185 parent_map[(gff_info['chr'], gff_info['info']['source'], gff_info['id'])] = dict(
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
186 type = gff_info['type'],
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
187 location = gff_info['location'],
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
188 strand = gff_info['strand'],
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
189 score = gff_info['score'],
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
190 name = tags.get('Name', [''])[0])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
191 elif rec_category == 'record':
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
192 #TODO how to handle plain records?
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
193 c = 1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
194 ga_handle.close()
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
195
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
196 # depends on file type create parent feature
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
197 if not ftype:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
198 parent_map, child_map = create_missing_feature_type(parent_map, child_map)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
199
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
200 # connecting parent child relations
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
201 # essentially the parent child features are here from any type of GTF/GFF2/GFF3 file
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
202 gene_mat = format_gene_models(parent_map, child_map)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
203
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
204 return gene_mat
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
205
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
206 def format_gene_models(parent_nf_map, child_nf_map):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
207 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
208 Genarate GeneObject based on the parsed file contents
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
209
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
210 @args parent_nf_map: parent features with source and chromosome information
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
211 @type parent_nf_map: collections defaultdict
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
212 @args child_nf_map: transctipt and exon information are encoded
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
213 @type child_nf_map: collections defaultdict
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
214 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
215
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
216 g_cnt = 0
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
217 gene = np.zeros((len(parent_nf_map),), dtype = utils.init_gene())
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
218
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
219 for pkey, pdet in parent_nf_map.items():
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
220 # considering only gene features
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
221 #if not re.search(r'gene', pdet.get('type', '')):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
222 # continue
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
223
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
224 # infer the gene start and stop if not there in the
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
225 if not pdet.get('location', []):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
226 GNS, GNE = [], []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
227 # multiple number of transcripts
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
228 for L1 in child_nf_map[pkey]:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
229 GNS.append(L1.get('location', [])[0])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
230 GNE.append(L1.get('location', [])[1])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
231 GNS.sort()
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
232 GNE.sort()
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
233 pdet['location'] = [GNS[0], GNE[-1]]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
234
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
235 orient = pdet.get('strand', '')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
236 gene[g_cnt]['id'] = g_cnt +1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
237 gene[g_cnt]['chr'] = pkey[0]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
238 gene[g_cnt]['source'] = pkey[1]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
239 gene[g_cnt]['name'] = pkey[-1]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
240 gene[g_cnt]['start'] = pdet.get('location', [])[0]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
241 gene[g_cnt]['stop'] = pdet.get('location', [])[1]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
242 gene[g_cnt]['strand'] = orient
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
243 gene[g_cnt]['score'] = pdet.get('score','')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
244
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
245 # default value
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
246 gene[g_cnt]['is_alt_spliced'] = gene[g_cnt]['is_alt'] = 0
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
247 if len(child_nf_map[pkey]) > 1:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
248 gene[g_cnt]['is_alt_spliced'] = gene[g_cnt]['is_alt'] = 1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
249
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
250 # complete sub-feature for all transcripts
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
251 dim = len(child_nf_map[pkey])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
252 TRS = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
253 TR_TYP = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
254 EXON = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
255 UTR5 = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
256 UTR3 = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
257 CDS = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
258 TISc = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
259 TSSc = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
260 CLV = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
261 CSTOP = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
262 TSTAT = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
263 TSCORE = np.zeros((dim,), dtype=np.object)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
264
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
265 # fetching corresponding transcripts
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
266 for xq, Lv1 in enumerate(child_nf_map[pkey]):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
267
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
268 TID = Lv1.get('ID', '')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
269 TRS[xq]= np.array([TID])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
270
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
271 TYPE = Lv1.get('type', '')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
272 TR_TYP[xq] = np.array('')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
273 TR_TYP[xq] = np.array(TYPE) if TYPE else TR_TYP[xq]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
274
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
275 orient = Lv1.get('strand', '')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
276 tr_score = Lv1.get('score', '')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
277
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
278 # fetching different sub-features
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
279 child_feat = defaultdict(list)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
280 for Lv2 in child_nf_map[(pkey[0], pkey[1], TID)]:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
281 E_TYP = Lv2.get('type', '')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
282 child_feat[E_TYP].append(Lv2.get('location'))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
283
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
284 # make general ascending order of coordinates
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
285 if orient == '-':
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
286 for etype, excod in child_feat.items():
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
287 if len(excod) > 1:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
288 if excod[0][0] > excod[-1][0]:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
289 excod.reverse()
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
290 child_feat[etype] = excod
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
291
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
292 # make exon coordinate from cds and utr regions
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
293 if not child_feat.get('exon'):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
294 if child_feat.get('CDS'):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
295 exon_cod = utils.make_Exon_cod( orient,
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
296 NonetoemptyList(child_feat.get('five_prime_UTR')),
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
297 NonetoemptyList(child_feat.get('CDS')),
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
298 NonetoemptyList(child_feat.get('three_prime_UTR')))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
299 child_feat['exon'] = exon_cod
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
300 else:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
301 # TODO only UTR's
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
302 # searching through keys to find a pattern describing exon feature
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
303 ex_key_pattern = [k for k in child_feat if k.endswith("exon")]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
304 if ex_key_pattern:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
305 child_feat['exon'] = child_feat[ex_key_pattern[0]]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
306
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
307 # stop_codon are seperated from CDS, add the coordinates based on strand
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
308 if child_feat.get('stop_codon'):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
309 if orient == '+':
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
310 if child_feat.get('stop_codon')[0][0] - child_feat.get('CDS')[-1][1] == 1:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
311 child_feat['CDS'][-1] = [child_feat.get('CDS')[-1][0], child_feat.get('stop_codon')[0][1]]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
312 else:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
313 child_feat['CDS'].append(child_feat.get('stop_codon')[0])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
314 elif orient == '-':
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
315 if child_feat.get('CDS')[0][0] - child_feat.get('stop_codon')[0][1] == 1:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
316 child_feat['CDS'][0] = [child_feat.get('stop_codon')[0][0], child_feat.get('CDS')[0][1]]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
317 else:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
318 child_feat['CDS'].insert(0, child_feat.get('stop_codon')[0])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
319
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
320 # transcript signal sites
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
321 TIS, cdsStop, TSS, cleave = [], [], [], []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
322 cds_status, exon_status, utr_status = 0, 0, 0
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
323
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
324 if child_feat.get('exon'):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
325 TSS = [child_feat.get('exon')[-1][1]]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
326 TSS = [child_feat.get('exon')[0][0]] if orient == '+' else TSS
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
327 cleave = [child_feat.get('exon')[0][0]]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
328 cleave = [child_feat.get('exon')[-1][1]] if orient == '+' else cleave
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
329 exon_status = 1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
330
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
331 if child_feat.get('CDS'):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
332 if orient == '+':
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
333 TIS = [child_feat.get('CDS')[0][0]]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
334 cdsStop = [child_feat.get('CDS')[-1][1]-3]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
335 else:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
336 TIS = [child_feat.get('CDS')[-1][1]]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
337 cdsStop = [child_feat.get('CDS')[0][0]+3]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
338 cds_status = 1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
339 # cds phase calculation
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
340 child_feat['CDS'] = utils.add_CDS_phase(orient, child_feat.get('CDS'))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
341
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
342 # sub-feature status
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
343 if child_feat.get('three_prime_UTR') or child_feat.get('five_prime_UTR'):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
344 utr_status =1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
345
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
346 if utr_status == cds_status == exon_status == 1:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
347 t_status = 1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
348 else:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
349 t_status = 0
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
350
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
351 # add sub-feature # make array for export to different out
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
352 TSTAT[xq] = t_status
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
353 EXON[xq] = np.array(child_feat.get('exon'), np.float64)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
354 UTR5[xq] = np.array(NonetoemptyList(child_feat.get('five_prime_UTR')))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
355 UTR3[xq] = np.array(NonetoemptyList(child_feat.get('three_prime_UTR')))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
356 CDS[xq] = np.array(NonetoemptyList(child_feat.get('CDS')))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
357 TISc[xq] = np.array(TIS)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
358 CSTOP[xq] = np.array(cdsStop)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
359 TSSc[xq] = np.array(TSS)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
360 CLV[xq] = np.array(cleave)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
361 TSCORE[xq] = tr_score
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
362
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
363 # add sub-features to the parent gene feature
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
364 gene[g_cnt]['transcript_status'] = TSTAT
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
365 gene[g_cnt]['transcripts'] = TRS
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
366 gene[g_cnt]['exons'] = EXON
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
367 gene[g_cnt]['utr5_exons'] = UTR5
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
368 gene[g_cnt]['cds_exons'] = CDS
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
369 gene[g_cnt]['utr3_exons'] = UTR3
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
370 gene[g_cnt]['transcript_type'] = TR_TYP
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
371 gene[g_cnt]['tis'] = TISc
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
372 gene[g_cnt]['cdsStop'] = CSTOP
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
373 gene[g_cnt]['tss'] = TSSc
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
374 gene[g_cnt]['cleave'] = CLV
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
375 gene[g_cnt]['transcript_score'] = TSCORE
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
376
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
377 gene[g_cnt]['gene_info'] = dict( ID = pkey[-1],
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
378 Name = pdet.get('name'),
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
379 Source = pkey[1])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
380 # few empty fields // TODO fill this:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
381 gene[g_cnt]['anno_id'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
382 gene[g_cnt]['confgenes_id'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
383 gene[g_cnt]['alias'] = ''
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
384 gene[g_cnt]['name2'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
385 gene[g_cnt]['chr_num'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
386 gene[g_cnt]['paralogs'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
387 gene[g_cnt]['transcript_valid'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
388 gene[g_cnt]['exons_confirmed'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
389 gene[g_cnt]['tis_conf'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
390 gene[g_cnt]['tis_info'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
391 gene[g_cnt]['cdsStop_conf'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
392 gene[g_cnt]['cdsStop_info'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
393 gene[g_cnt]['tss_info'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
394 gene[g_cnt]['tss_conf'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
395 gene[g_cnt]['cleave_info'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
396 gene[g_cnt]['cleave_conf'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
397 gene[g_cnt]['polya_info'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
398 gene[g_cnt]['polya_conf'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
399 gene[g_cnt]['is_valid'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
400 gene[g_cnt]['transcript_complete'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
401 gene[g_cnt]['is_complete'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
402 gene[g_cnt]['is_correctly_gff3_referenced'] = ''
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
403 gene[g_cnt]['splicegraph'] = []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
404 g_cnt += 1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
405
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
406 ## deleting empty gene records from the main array
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
407 XPFLG=0
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
408 for XP, ens in enumerate(gene):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
409 if ens[0]==0:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
410 XPFLG=1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
411 break
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
412
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
413 if XPFLG==1:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
414 XQC = range(XP, len(gene)+1)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
415 gene = np.delete(gene, XQC)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
416
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
417 return gene
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
418
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
419 def NonetoemptyList(XS):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
420 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
421 Convert a None type to empty list
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
422
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
423 @args XS: None type
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
424 @type XS: str
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
425 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
426 return [] if XS is None else XS
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
427
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
428 def create_missing_feature_type(p_feat, c_feat):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
429 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
430 GFF/GTF file defines only child features. This function tries to create
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
431 the parent feature from the information provided in the attribute column.
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
432
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
433 example:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
434 chr21 hg19_knownGene exon 9690071 9690100 0.000000 + . gene_id "uc002zkg.1"; transcript_id "uc002zkg.1";
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
435 chr21 hg19_knownGene exon 9692178 9692207 0.000000 + . gene_id "uc021wgt.1"; transcript_id "uc021wgt.1";
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
436 chr21 hg19_knownGene exon 9711935 9712038 0.000000 + . gene_id "uc011abu.2"; transcript_id "uc011abu.2";
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
437
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
438 This function gets the parsed feature annotations.
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
439
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
440 @args p_feat: Parent feature map
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
441 @type p_feat: collections defaultdict
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
442 @args c_feat: Child feature map
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
443 @type c_feat: collections defaultdict
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
444 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
445
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
446 child_n_map = defaultdict(list)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
447 for fid, det in c_feat.items():
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
448 # get the details from grand child
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
449 GID = STRD = SCR = None
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
450 SPOS, EPOS = [], []
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
451 TYP = dict()
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
452 for gchild in det:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
453 GID = gchild.get('gene_id', [''])[0]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
454 SPOS.append(gchild.get('location', [])[0])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
455 EPOS.append(gchild.get('location', [])[1])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
456 STRD = gchild.get('strand', '')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
457 SCR = gchild.get('score', '')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
458 if gchild.get('type', '') == "gene": ## gencode GTF file has this problem
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
459 continue
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
460 TYP[gchild.get('type', '')] = 1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
461 SPOS.sort()
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
462 EPOS.sort()
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
463
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
464 # infer transcript type
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
465 transcript_type = 'transcript'
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
466 transcript_type = 'mRNA' if TYP.get('CDS', '') or TYP.get('cds', '') else transcript_type
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
467
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
468 # gene id and transcript id are same
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
469 transcript_id = fid[-1]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
470 if GID == transcript_id:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
471 transcript_id = 'Transcript:' + str(GID)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
472
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
473 # level -1 feature type
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
474 p_feat[(fid[0], fid[1], GID)] = dict( type = 'gene',
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
475 location = [], ## infer location based on multiple transcripts
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
476 strand = STRD,
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
477 name = GID )
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
478 # level -2 feature type
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
479 child_n_map[(fid[0], fid[1], GID)].append(
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
480 dict( type = transcript_type,
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
481 location = [SPOS[0], EPOS[-1]],
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
482 strand = STRD,
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
483 score = SCR,
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
484 ID = transcript_id,
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
485 gene_id = '' ))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
486 # reorganizing the grand child
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
487 for gchild in det:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
488 child_n_map[(fid[0], fid[1], transcript_id)].append(
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
489 dict( type = gchild.get('type', ''),
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
490 location = gchild.get('location'),
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
491 strand = gchild.get('strand'),
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
492 ID = gchild.get('ID'),
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
493 score = gchild.get('score'),
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
494 gene_id = '' ))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
495 return p_feat, child_n_map
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
496