annotate BigBed.py @ 59:f311e21ef362 draft

planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 72268f816f10341dbe3f6b33f8d9170d859c9697-dirty
author yating-l
date Fri, 30 Jun 2017 15:50:57 -0400
parents b39dd0b5a166
children
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"]
57
b39dd0b5a166 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit ce28781f52a4f84039de300cb41e3982f2e8bf51-dirty
yating-l
parents: 55
diff changeset
24 self.database = data_bigbed["database"]
55
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
25
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
26 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
27 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
28 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
29 else:
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
30 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
31
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
32 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
33
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
34 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
35
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
36 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
37
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
38 # Create the Track Object
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
39 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
40 track_name=track_name,
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
41 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
42 track_type=track_type,
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
43 visibility='hide',
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
44 priority=self.priority,
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
45 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
46 track_color=self.track_color,
57
b39dd0b5a166 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit ce28781f52a4f84039de300cb41e3982f2e8bf51-dirty
yating-l
parents: 55
diff changeset
47 group_name=self.group_name,
b39dd0b5a166 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit ce28781f52a4f84039de300cb41e3982f2e8bf51-dirty
yating-l
parents: 55
diff changeset
48 database=self.database)
55
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 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
51
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 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
54 """
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
55 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
56
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
57 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
58 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
59
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
60 :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
61
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
62 :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
63 """
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
64
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
65 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
66 stdout=PIPE, bufsize=1)
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 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
69
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
70 with cmd_ph.stdout:
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
71 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
72 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
73
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
74 if match:
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
75 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
76 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
77 break
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 cmd_ph.wait()
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
80
ba9997c847dc planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit b2894a39921adaa6a93b12523820b61a78e11f20-dirty
yating-l
parents:
diff changeset
81 return bed_type