Mercurial > repos > rmarenco > hubarchivecreator
comparison Gtf.py @ 1:fb5e60d4d18a draft
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
| author | rmarenco |
|---|---|
| date | Wed, 13 Jul 2016 13:36:37 -0400 |
| parents | |
| children | acc233161f50 |
comparison
equal
deleted
inserted
replaced
| 0:0f3bc17e5ede | 1:fb5e60d4d18a |
|---|---|
| 1 #!/usr/bin/python | |
| 2 | |
| 3 import os | |
| 4 import tempfile | |
| 5 | |
| 6 # Internal dependencies | |
| 7 from Datatype import Datatype | |
| 8 from Track import Track | |
| 9 from TrackDb import TrackDb | |
| 10 from util import subtools | |
| 11 | |
| 12 | |
| 13 class Gtf( Datatype ): | |
| 14 def __init__( self, input_gtf_false_path, data_gtf, | |
| 15 input_fasta_file, extra_files_path, tool_directory ): | |
| 16 super(Gtf, self).__init__( input_fasta_file=input_fasta_file, | |
| 17 extra_files_path=extra_files_path, | |
| 18 tool_directory=tool_directory ) | |
| 19 | |
| 20 self.track = None | |
| 21 | |
| 22 self.input_gtf_false_path = input_gtf_false_path | |
| 23 self.name_gtf = data_gtf["name"] | |
| 24 self.priority = data_gtf["order_index"] | |
| 25 | |
| 26 print "Creating TrackHub GTF from (falsePath: %s; name: %s)" % ( self.input_gtf_false_path, self.name_gtf) | |
| 27 | |
| 28 # TODO: See if we need these temporary files as part of the generated files | |
| 29 genePredFile = tempfile.NamedTemporaryFile(bufsize=0, suffix=".genePred") | |
| 30 unsortedBedFile = tempfile.NamedTemporaryFile(bufsize=0, suffix=".unsortedBed") | |
| 31 sortedBedFile = tempfile.NamedTemporaryFile(suffix=".sortedBed") | |
| 32 | |
| 33 twoBitInfoFile = tempfile.NamedTemporaryFile(bufsize=0) | |
| 34 chromSizesFile = tempfile.NamedTemporaryFile(bufsize=0, suffix=".chrom.sizes") | |
| 35 | |
| 36 # GtfToGenePred | |
| 37 subtools.gtfToGenePred(self.input_gtf_false_path, genePredFile.name) | |
| 38 | |
| 39 # TODO: From there, refactor because common use with Gff3.py | |
| 40 # genePredToBed processing | |
| 41 subtools.genePredToBed(genePredFile.name, unsortedBedFile.name) | |
| 42 | |
| 43 # Sort processing | |
| 44 subtools.sort(unsortedBedFile.name, sortedBedFile.name) | |
| 45 | |
| 46 # TODO: Chehck if the twoBitInfo / ChromSizes is redundant and make an intermediate class | |
| 47 # Generate the twoBitInfo | |
| 48 subtools.twoBitInfo(self.twoBitFile.name, twoBitInfoFile.name) | |
| 49 | |
| 50 # Then we get the output to generate the chromSizes | |
| 51 # TODO: Check if no errors | |
| 52 subtools.sortChromSizes(twoBitInfoFile.name, chromSizesFile.name) | |
| 53 | |
| 54 # bedToBigBed processing | |
| 55 # TODO: Change the name of the bb, to tool + genome + possible adding if multiple + .bb | |
| 56 trackName = "".join( ( self.name_gtf, ".bb") ) | |
| 57 myBigBedFilePath = os.path.join(self.myTrackFolderPath, trackName) | |
| 58 with open(myBigBedFilePath, 'w') as bigBedFile: | |
| 59 subtools.bedToBigBed(sortedBedFile.name, chromSizesFile.name, bigBedFile.name) | |
| 60 | |
| 61 # Create the Track Object | |
| 62 dataURL = "tracks/%s" % trackName | |
| 63 | |
| 64 trackDb = TrackDb( | |
| 65 trackName=trackName, | |
| 66 longLabel=self.name_gtf, | |
| 67 shortLabel=self.getShortName( self.name_gtf ), | |
| 68 trackDataURL=dataURL, | |
| 69 trackType='bigBed 12 +', | |
| 70 visibility='dense', | |
| 71 priority=self.priority, | |
| 72 ) | |
| 73 self.track = Track( | |
| 74 trackFile=myBigBedFilePath, | |
| 75 trackDb=trackDb, | |
| 76 ) | |
| 77 | |
| 78 print("- %s created in %s" % (trackName, myBigBedFilePath)) |