Mercurial > repos > rmarenco > hubarchivecreator

annotate Gff3.py @ 9:4f9847539a28 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 1b241bbccfc1b90ce2626dc4f49788bb9e693a80
author rmarenco
date Wed, 20 Jul 2016 12:29:08 -0400
parents fb5e60d4d18a
children acc233161f50
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 Gff3( Datatype ):
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
14 def __init__( self, input_Gff3_false_path, data_gff3,
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
15 input_fasta_false_path, extra_files_path, tool_directory ):
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
16 super( Gff3, self ).__init__(
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
17 input_fasta_false_path, extra_files_path, tool_directory
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
18 )
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.track = None
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
21
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
22 self.input_Gff3_false_path = input_Gff3_false_path
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
23 self.name_gff3 = data_gff3["name"]
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
24 self.priority = data_gff3["order_index"]
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 # TODO: Refactor into another Class to manage the twoBitInfo and ChromSizes (same process as in Gtf.py)
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
32 twoBitInfoFile = tempfile.NamedTemporaryFile(bufsize=0)
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
33 chromSizesFile = tempfile.NamedTemporaryFile(bufsize=0, suffix=".chrom.sizes")
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
34
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
35 # gff3ToGenePred processing
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
36 subtools.gff3ToGenePred(self.input_Gff3_false_path, genePredFile.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 # TODO: From there, refactor because common use with Gtf.py
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
39 # genePredToBed processing
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
40 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
41
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
42 # Sort processing
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
43 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
44
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
45 # Generate the twoBitInfo
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
46 subtools.twoBitInfo(self.twoBitFile.name, twoBitInfoFile.name)
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 # Then we get the output to generate the chromSizes
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
49 # TODO: Check if no errors
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
50 subtools.sortChromSizes(twoBitInfoFile.name, chromSizesFile.name)
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
51
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
52 # bedToBigBed processing
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
53 # 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
54 trackName = "".join( (self.name_gff3, ".bb" ) )
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
55 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
56 with open(myBigBedFilePath, 'w') as bigBedFile:
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
57 subtools.bedToBigBed(sortedBedFile.name, chromSizesFile.name, bigBedFile.name)
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
58
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
59 # Create the Track Object
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
60 dataURL = "tracks/%s" % trackName
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
61
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
62 trackDb = TrackDb(
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
63 trackName=trackName,
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
64 longLabel=self.name_gff3,
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
65 shortLabel=self.getShortName( self.name_gff3 ),
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
66 trackDataURL=dataURL,
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
67 trackType='bigBed 12 +',
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
68 visibility='dense',
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
69 priority=self.priority,
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
70 )
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
71
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
72 self.track = Track(
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
73 trackFile=myBigBedFilePath,
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
74 trackDb=trackDb,
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
75 )
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
76
fb5e60d4d18a planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff changeset
77 print("- %s created in %s" % (trackName, myBigBedFilePath))