annotate BigBed.py @ 55:ba9997c847dc draft

planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
author yating-l
date Wed, 17 May 2017 13:16:57 -0400
parents
children b39dd0b5a166
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
55
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
1 #!/usr/bin/python
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
2
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
3 import os
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
4 import shutil
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
5 from subprocess import Popen, PIPE
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
6 import re
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
7
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
8 # Internal dependencies
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
9 from Datatype import Datatype
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
10
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
11 class BigBed(Datatype):
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
12 """ Configurations for creating the bigBed evidence track """
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
13
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
14 def __init__(self, input_bigbed_path, data_bigbed):
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
15 super(BigBed, self).__init__()
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
16
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
17 self.track = None
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
18
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
19 self.input_bigbed_path = input_bigbed_path
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
20 self.name_bigbed = data_bigbed["name"]
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
21 self.priority = data_bigbed["order_index"]
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
22 self.track_color = data_bigbed["track_color"]
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
23 self.group_name = data_bigbed["group_name"]
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
24
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
25 track_name = "".join((self.name_bigbed, ".bigbed"))
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
26 if data_bigbed["long_label"]:
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
27 self.long_label = data_bigbed["long_label"]
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
28 else:
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
29 self.long_label = self.name_bigbed
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
30
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
31 bigbed_file_path = os.path.join(self.myTrackFolderPath, track_name)
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
32
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
33 track_type = self.determine_track_type(input_bigbed_path)
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
34
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
35 shutil.copy(self.input_bigbed_path, bigbed_file_path)
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
36
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
37 # Create the Track Object
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
38 self.createTrack(file_path=track_name,
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
39 track_name=track_name,
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
40 long_label=self.long_label,
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
41 track_type=track_type,
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
42 visibility='hide',
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
43 priority=self.priority,
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
44 track_file=bigbed_file_path,
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
45 track_color=self.track_color,
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
46 group_name=self.group_name)
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
47
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
48 print "- BigBed %s created" % self.name_bigbed
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
49
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
50
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
51 def determine_track_type(self, bb_file):
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
52 """
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
53 Determine the number of standard and extra fields using bigBedSummary
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
54
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
55 Implementation of reading from stdout is based on a Stackoverflow post:
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
56 http://stackoverflow.com/questions/2715847/python-read-streaming-input-from-subprocess-communicate
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
57
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
58 :param bb_file: path to a bigBed file
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
59
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
60 :returns: the bigBed track type
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
61 """
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
62
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
63 cmd_ph = Popen(["bigBedSummary", "-fields", bb_file, "stdout"],
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
64 stdout=PIPE, bufsize=1)
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
65
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
66 pattern = r"(\d+) bed definition fields, (\d+) total fields"
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
67
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
68 with cmd_ph.stdout:
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
69 for line in iter(cmd_ph.stdout.readline, b''):
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
70 match = re.match(pattern, line)
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
71
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
72 if match:
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
73 extra_mark = "." if match.group(1) == match.group(2) else "+"
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
74 bed_type = "bigBed %s %s" % (match.group(1), extra_mark)
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
75 break
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
76
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
77 cmd_ph.wait()
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
78
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
79 return bed_type