Mercurial > repos > yating-l > jbrowsearchivecreator

annotate utils.py @ 3:eda851e52060 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 479fc6490e24ca0e5034ae6c3579882e97e095e6-dirty
author yating-l
date Wed, 31 May 2017 15:45:47 -0400
parents 804a93e87cc8
children 7e471cdd9e71
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
1 #!/usr/bin/env python
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
2
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
3 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
4 This file include common used functions for converting file format to gff3
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
5 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
6 from collections import OrderedDict
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
7 import json
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
8 import subprocess
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
9 import os
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
10 import tempfile
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
11 import string
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
12
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
13 def write_features(field, attribute, gff3):
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
14 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
15 The function write the features to gff3 format (defined in https://github.com/The-Sequence-Ontology/Specifications/blob/master/gff3.md)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
16 field, attribute are ordered dictionary
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
17 gff3 is the file handler
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
18 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
19 attr = []
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
20 for v in field.values():
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
21 gff3.write(str(v) + '\t')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
22 for k, v in attribute.items():
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
23 s = str(k) + '=' + str(v)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
24 attr.append(s)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
25 gff3.write(';'.join(attr))
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
26 gff3.write('\n')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
27
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
28 def getChromSizes(reference, tool_dir):
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
29 #TODO: find a better way instead of shipping the two exec files with the tool
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
30 faToTwoBit = os.path.join(tool_dir, 'faToTwoBit')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
31 twoBitInfo = os.path.join(tool_dir, 'twoBitInfo')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
32 try:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
33 twoBitFile = tempfile.NamedTemporaryFile(bufsize=0)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
34 chrom_sizes = tempfile.NamedTemporaryFile(bufsize=0, suffix='.chrom.sizes', delete=False)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
35 except IOError as err:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
36 print "Cannot create tempfile err({0}): {1}".format(err.errno, err.strerror)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
37 try:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
38 subprocess.call(['faToTwoBit', reference, twoBitFile.name])
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
39 except OSError as err:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
40 print "Cannot generate twoBitFile from faToTwoBit err({0}): {1}".format(err.errno, err.strerror)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
41 try:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
42 subprocess.call(['twoBitInfo', twoBitFile.name, chrom_sizes.name])
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
43 except OSError as err:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
44 print "Cannot generate chrom_sizes from twoBitInfo err({0}): {1}".format(err.errno, err.strerror)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
45 return chrom_sizes
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
46
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
47 def sequence_region(chrom_sizes):
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
48 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
49 This function read from a chromatin size file generated by twoBitInfo and write the information to dict
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
50 return a dict
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
51 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
52 f = open(chrom_sizes, 'r')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
53 sizes = f.readlines()
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
54 sizes_dict = {}
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
55 for line in sizes:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
56 chrom_info = line.rstrip().split('\t')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
57 sizes_dict[chrom_info[0]] = chrom_info[1]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
58 return sizes_dict
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
59
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
60 def child_blocks(parent_field, parent_attr, gff3):
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
61 num = 0
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
62 blockcount = int(parent_attr['blockcount'])
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
63 chromstart = parent_attr['chromstarts'].split(',')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
64 blocksize = parent_attr['blocksizes'].split(',')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
65 while num < blockcount:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
66 child_attr = OrderedDict()
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
67 child_field = parent_field
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
68 child_field['type'] = 'exon_junction'
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
69 child_field['start'] = int(chromstart[num]) + int(parent_field['start'])
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
70 child_field['end'] = int(child_field['start']) + int(blocksize[num]) - 1
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
71 child_attr['ID'] = parent_attr['ID'] + '_exon_' + str(num+1)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
72 child_attr['Parent'] = parent_attr['ID']
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
73 write_features(child_field, child_attr, gff3)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
74 num = num + 1
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
75
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
76 def add_tracks_to_json(trackList_json, new_tracks, modify_type):
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
77 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
78 Add to track configuration (trackList.json)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
79 # modify_type = 'add_tracks': add a new track like bam or bigwig, new_track = dict()
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
80 # modify_type = 'add_attr': add configuration to the existing track, new_track = dict(track_name: dict())
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
81 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
82 with open(trackList_json, 'r+') as f:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
83 data = json.load(f)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
84 if modify_type == 'add_tracks':
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
85 data['tracks'].append(new_tracks)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
86 elif modify_type == 'add_attr':
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
87 for k in new_tracks:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
88 for track in data['tracks']:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
89 if k.lower() in track['urlTemplate'].lower():
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
90 attr = new_tracks[k]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
91 for k, v in attr.items():
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
92 track[k] = v
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
93 f.seek(0, 0)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
94 f.write(json.dumps(data, separators=(',' , ':'), indent=4))
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
95 f.truncate()
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
96 f.close()
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
97
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
98 def gtfToGff3(gtf_file, gff3_file, chrom_sizes):
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
99 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
100 Covert gtf file output from StringTie to gff3 format
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
101 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
102 gff3 = open(gff3_file, 'w')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
103 gff3.write("##gff-version 3\n")
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
104 sizes_dict = sequence_region(chrom_sizes)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
105 seq_regions = dict()
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
106 parents = dict()
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
107 with open(gtf_file, 'r') as gtf:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
108 for line in gtf:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
109 if line.startswith('#'):
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
110 continue
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
111 field = OrderedDict()
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
112 attribute = OrderedDict()
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
113 li = line.rstrip().split("\t")
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
114 #print li
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
115 field['seqid'] = li[0]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
116 #print field['seqid']
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
117 if field['seqid'] not in seq_regions:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
118 end_region = sizes_dict[field['seqid']]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
119 gff3.write("##sequence-region " + field['seqid'] + ' 1 ' + str(end_region) + '\n')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
120 seq_regions[field['seqid']] = end_region
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
121 field['source'] = li[1]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
122 field['type'] = li[2]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
123 # The first base in a chromosome is numbered 0 in BED format
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
124 field['start'] = li[3]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
125 field['end'] = li[4]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
126 field['score'] = li[5]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
127 field['strand'] = li[6]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
128 field['phase'] = li[7]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
129 attr_li = li[8].split(';')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
130 gene_id = attr_li[0].split()[1].strip('"')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
131 attribute['ID'] = gene_id + '_' + field['type'] + '_' + str(field['start']) + '_' + str(field['end'])
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
132 if field['type'] == 'transcript':
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
133 parents[gene_id] = attribute['ID']
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
134 attribute['transcript_id'] = attr_li[1].split()[1].strip('"')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
135 attribute['coverage'] = attr_li[2].split()[1].strip('"')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
136 attribute['fpkm'] = attr_li[3].split()[1].strip('"')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
137 attribute['tpm'] = attr_li[4].split()[1].strip('"')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
138 elif field['type'] == 'exon':
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
139 attribute['Parent'] = parents[gene_id]
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
140 attribute['transcript_id'] = attr_li[1].split()[1].strip('"')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
141 attribute['coverage'] = attr_li[3].split()[1].strip('"')
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
142 write_features(field, attribute, gff3)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
143 gff3.close()
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
144
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
145
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
146 def sanitize_name(input_name):
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
147 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
148 Galaxy will name all the files and dirs as *.dat,
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
149 the function can replace '.' to '_' for the dirs
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
150 """
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
151 validChars = "_-%s%s" % (string.ascii_letters, string.digits)
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
152 sanitized_name = ''.join([c if c in validChars else '_' for c in input_name])
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
153 return "gonramp_" + sanitized_name
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
154
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
155 def createBamIndex(bamfile):
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
156 subprocess.call(['samtools', 'index', bamfile])
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
157 filename = bamfile + '.bai'
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
158 if os.path.exists(filename):
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
159 return filename
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
160 else:
804a93e87cc8 planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit f22711ea7a464bdaf4d5aaea07f2eacf967aa66e-dirty
yating-l
parents:
diff changeset
161 raise ValueError('Did not find bai file')