Mercurial > repos > yating-l > jbrowsearchivecreator

annotate bedToGff3.py @ 9:ee187d1e0798 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
author yating-l
date Fri, 19 May 2017 16:32:23 -0400
parents 8d1cf7ce65cd
children 671231da45f9
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
1 #!/usr/bin/env python
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
2
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
3 '''
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
4 Convert BED format to gff3
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
5 '''
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
6 import os
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
7 from collections import OrderedDict
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
8 import utils
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
9
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
10 class bedToGff3():
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
11 def __init__(self, inputBedFile, chrom_sizes, bed_type, output):
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
12 self.input = inputBedFile
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
13 #file_dir = os.path.basename(inputBedFile)
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
14 #print file_dir + "\n\n"
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
15 self.output = output
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
16 self.chrom_sizes = chrom_sizes
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
17 self.type = bed_type
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
18 if self.type == "trfbig":
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
19 self.trfbig_to_gff3()
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
20 if self.type == "regtools":
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
21 self.splicejunctions_to_gff3()
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
22
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
23 def trfbig_to_gff3(self):
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
24 gff3 = open(self.output, 'w')
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
25 gff3.write("##gff-version 3\n")
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
26 sizes_dict = utils.sequence_region(self.chrom_sizes)
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
27 seq_regions = dict()
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
28 with open(self.input, 'r') as bed:
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
29 for line in bed:
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
30 field = OrderedDict()
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
31 attribute = OrderedDict()
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
32 li = line.rstrip().split("\t")
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
33 field['seqid'] = li[0]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
34 if field['seqid'] not in seq_regions:
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
35 end_region = sizes_dict[field['seqid']]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
36 gff3.write("##sequence-region " + field['seqid'] + ' 1 ' + str(end_region) + '\n')
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
37 seq_regions[field['seqid']] = end_region
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
38 field['source'] = li[3]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
39 field['type'] = 'tandem_repeat'
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
40 # The first base in a chromosome is numbered 0 in BED format
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
41 field['start'] = str(int(li[1]) + 1)
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
42 field['end'] = li[2]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
43 field['score'] = li[9]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
44 field['strand'] = '+'
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
45 field['phase'] = '.'
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
46 attribute['length of repeat unit'] = li[4]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
47 attribute['mean number of copies of repeat'] = li[5]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
48 attribute['length of consensus sequence'] = li[6]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
49 attribute['percentage match'] = li[7]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
50 attribute['percentage indel'] = li[8]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
51 attribute['percent of a\'s in repeat unit'] = li[10]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
52 attribute['percent of c\'s in repeat unit'] = li[11]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
53 attribute['percent of g\'s in repeat unit'] = li[12]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
54 attribute['percent of t\'s in repeat unit'] = li[13]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
55 attribute['entropy'] = li[14]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
56 attribute['sequence of repeat unit element'] = li[15]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
57 utils.write_features(field, attribute, gff3)
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
58 gff3.close()
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
59
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
60
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
61 def splicejunctions_to_gff3(self):
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
62 gff3 = open(self.output, 'w')
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
63 gff3.write("##gff-version 3\n")
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
64 sizes_dict = utils.sequence_region(self.chrom_sizes)
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
65 seq_regions = dict()
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
66 with open(self.input, 'r') as bed:
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
67 for line in bed:
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
68 field = OrderedDict()
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
69 attribute = OrderedDict()
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
70 li = line.rstrip().split("\t")
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
71 field['seqid'] = li[0]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
72 if field['seqid'] not in seq_regions:
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
73 end_region = sizes_dict[field['seqid']]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
74 gff3.write("##sequence-region " + field['seqid'] + ' 1 ' + str(end_region) + '\n')
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
75 seq_regions[field['seqid']] = end_region
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
76 field['source'] = li[3]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
77 field['type'] = 'junction'
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
78 # The first base in a chromosome is numbered 0 in BED format
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
79 field['start'] = int(li[1]) + 1
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
80 field['end'] = li[2]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
81 field['score'] = li[12]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
82 field['strand'] = li[5]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
83 field['phase'] = '.'
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
84 attribute['ID'] = li[3]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
85 attribute['Name'] = li[3]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
86 attribute['blockcount'] = li[9]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
87 attribute['blocksizes'] = li[10]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
88 attribute['chromstarts'] = li[11]
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
89 utils.write_features(field, attribute, gff3)
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
90 utils.child_blocks(field, attribute, gff3)
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
91 gff3.close()
8d1cf7ce65cd planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit d583ac16a6c6942730ea536eb59cc37941816030-dirty
yating-l
parents:
diff changeset
92