Mercurial > repos > rmarenco > hubarchivecreator
annotate Gtf.py @ 11:d05236b15f81 draft
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
author | rmarenco |
---|---|
date | Wed, 27 Jul 2016 10:10:49 -0400 |
parents | acc233161f50 |
children | 25809f699cb3 |
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"] |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
23 |
11
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
24 #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
|
25 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
26 # 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
|
27 genePredFile = tempfile.NamedTemporaryFile(bufsize=0, suffix=".genePred") |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
28 unsortedBedFile = tempfile.NamedTemporaryFile(bufsize=0, suffix=".unsortedBed") |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
29 sortedBedFile = tempfile.NamedTemporaryFile(suffix=".sortedBed") |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
30 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
31 # GtfToGenePred |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
32 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
|
33 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
34 # TODO: From there, refactor because common use with Gff3.py |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
35 # genePredToBed processing |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
36 subtools.genePredToBed(genePredFile.name, unsortedBedFile.name) |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
37 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
38 # Sort processing |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
39 subtools.sort(unsortedBedFile.name, sortedBedFile.name) |
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 # bedToBigBed processing |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
42 # TODO: Change the name of the bb, to tool + genome + possible adding if multiple + .bb |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
43 trackName = "".join( ( self.name_gtf, ".bb") ) |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
44 myBigBedFilePath = os.path.join(self.myTrackFolderPath, trackName) |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
45 with open(myBigBedFilePath, 'w') as bigBedFile: |
10
acc233161f50
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1b1063f90004764bcf504f4340738eca5c4b1f9d
rmarenco
parents:
1
diff
changeset
|
46 subtools.bedToBigBed(sortedBedFile.name, self.chromSizesFile.name, bigBedFile.name) |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
47 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
48 # Create the Track Object |
11
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
49 self.createTrack(file_path=trackName, |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
50 track_name=trackName, |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
51 long_label=self.name_gtf, track_type='bigBed 12 +', visibility='dense', priority=self.priority, |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
52 track_file=myBigBedFilePath) |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
53 # |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
54 # dataURL = "tracks/%s" % trackName |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
55 # |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
56 # trackDb = TrackDb( |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
57 # trackName=trackName, |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
58 # longLabel=self.name_gtf, |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
59 # shortLabel=self.getShortName( self.name_gtf ), |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
60 # trackDataURL=dataURL, |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
61 # trackType='bigBed 12 +', |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
62 # visibility='dense', |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
63 # priority=self.priority, |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
64 # ) |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
65 # self.track = Track( |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
66 # trackFile=myBigBedFilePath, |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
67 # trackDb=trackDb, |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
68 # ) |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
69 |
11
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
70 print("- Gtf %s created" % self.name_gtf) |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
71 #print("- %s created in %s" % (trackName, myBigBedFilePath)) |