Mercurial > repos > rmarenco > hubarchivecreator

annotate Gtf.py @ 17:c02720d1afee draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
author rmarenco
date Fri, 30 Sep 2016 15:14:24 -0400
parents 3233451a3bd6
children 0152500d9acd
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
1 #!/usr/bin/python
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
2
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
3 import os
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
4 import tempfile
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
5
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
6 # Internal dependencies
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
7 from Datatype import Datatype
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
8 from Track import Track
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
9 from TrackDb import TrackDb
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
10 from util import subtools
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
11
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
12
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
13 class Gtf( Datatype ):
10
acc233161f50 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1b1063f90004764bcf504f4340738eca5c4b1f9d
rmarenco
parents: 1
diff changeset
14 def __init__( self, input_gtf_false_path, data_gtf):
acc233161f50 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1b1063f90004764bcf504f4340738eca5c4b1f9d
rmarenco
parents: 1
diff changeset
15
acc233161f50 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1b1063f90004764bcf504f4340738eca5c4b1f9d
rmarenco
parents: 1
diff changeset
16 super(Gtf, self).__init__()
1
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
17
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
18 self.track = None
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
19
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
20 self.input_gtf_false_path = input_gtf_false_path
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
21 self.name_gtf = data_gtf["name"]
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
22 self.priority = data_gtf["order_index"]
16
3233451a3bd6 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit fc73ec22a0db3ab09c4ac13dc58f0b54ae37845c
rmarenco
parents: 13
diff changeset
23 self.track_color = data_gtf["track_color"]
17
c02720d1afee planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
rmarenco
parents: 16
diff changeset
24 # TODO: Think about how to avoid repetition of the group_name everywhere
c02720d1afee planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
rmarenco
parents: 16
diff changeset
25 self.group_name = data_gtf["group_name"]
1
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
26
11
d05236b15f81 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents: 10
diff changeset
27 #print "Creating TrackHub GTF from (falsePath: %s; name: %s)" % ( self.input_gtf_false_path, self.name_gtf)
1
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
28
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
29 # TODO: See if we need these temporary files as part of the generated files
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
30 genePredFile = tempfile.NamedTemporaryFile(bufsize=0, suffix=".genePred")
13
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
31 unsorted_bigGenePred_file = tempfile.NamedTemporaryFile(bufsize=0, suffix=".unsorted.bigGenePred")
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
32 sorted_bigGenePred_file = tempfile.NamedTemporaryFile(suffix=".sortedBed.bigGenePred")
1
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
33
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
34 # GtfToGenePred
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
35 subtools.gtfToGenePred(self.input_gtf_false_path, genePredFile.name)
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
36
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
37 # TODO: From there, refactor because common use with Gff3.py
13
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
38 # genePredToBigGenePred processing
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
39 subtools.genePredToBigGenePred(genePredFile.name, unsorted_bigGenePred_file.name)
1
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
40
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
41 # Sort processing
13
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
42 subtools.sort(unsorted_bigGenePred_file.name, sorted_bigGenePred_file.name)
1
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
43
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
44 # bedToBigBed processing
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
45 trackName = "".join( ( self.name_gtf, ".bb") )
13
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
46
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
47 auto_sql_option = os.path.join(self.tool_directory, 'bigGenePred.as')
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
48
1
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
49 myBigBedFilePath = os.path.join(self.myTrackFolderPath, trackName)
13
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
50
1
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
51 with open(myBigBedFilePath, 'w') as bigBedFile:
13
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
52 subtools.bedToBigBed(sorted_bigGenePred_file.name,
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
53 self.chromSizesFile.name,
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
54 bigBedFile.name,
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
55 autoSql=auto_sql_option,
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
56 typeOption='bed12+8',
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
57 tab=True)
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
58
1
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
59
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
60 # Create the Track Object
11
d05236b15f81 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents: 10
diff changeset
61 self.createTrack(file_path=trackName,
d05236b15f81 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents: 10
diff changeset
62 track_name=trackName,
13
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
63 long_label=self.name_gtf, track_type='bigGenePred',
25809f699cb3 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents: 11
diff changeset
64 visibility='dense', priority=self.priority,
16
3233451a3bd6 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit fc73ec22a0db3ab09c4ac13dc58f0b54ae37845c
rmarenco
parents: 13
diff changeset
65 track_file=myBigBedFilePath,
17
c02720d1afee planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
rmarenco
parents: 16
diff changeset
66 track_color=self.track_color,
c02720d1afee planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
rmarenco
parents: 16
diff changeset
67 group_name=self.group_name)
1
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
68
11
d05236b15f81 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents: 10
diff changeset
69 print("- Gtf %s created" % self.name_gtf)