Mercurial > repos > yating-l > jbrowse_hub
changeset 50:3e5b6f6a9abe draft
planemo upload for repository https://github.com/Yating-L/jbrowse_hub commit 6391cdb975d14b5a7060fa1665a8a575797a0eaa-dirty
| author | yating-l |
|---|---|
| date | Thu, 30 Mar 2017 16:03:16 -0400 |
| parents | 193e94a43f0f |
| children | acedf82ea46c |
| files | TrackHub.py jbrowse_hub.py jbrowse_hub.xml trackObject.py utils.py |
| diffstat | 5 files changed, 75 insertions(+), 35 deletions(-) [+] |
line wrap: on
line diff
--- a/TrackHub.py Wed Mar 29 14:46:23 2017 -0400 +++ b/TrackHub.py Thu Mar 30 16:03:16 2017 -0400 @@ -3,6 +3,7 @@ import os import subprocess import shutil +import json import utils @@ -38,9 +39,7 @@ def prepareRefseq(self): try: #print os.path.join(self.tool_dir, 'prepare-refseqs.pl') + ", '--fasta', " + self.reference +", '--out', self.json])" - p = subprocess.Popen(['prepare-refseqs.pl', '--fasta', self.reference, '--out', self.json]) - # Wait for process to terminate. - p.communicate() + subprocess.call(['prepare-refseqs.pl', '--fasta', self.reference, '--out', self.json]) except OSError as e: print "Cannot prepare reference error({0}): {1}".format(e.errno, e.strerror) #TODO: hard coded the bam and bigwig tracks. Need to allow users to customize the settings @@ -57,18 +56,18 @@ elif track['dataType'] == 'bigwig': self.BigWig(track, metadata) else: - gff3_file = os.path.join(self.raw, track['fileName']) - if track['dataType'] == 'bedSpliceJunctions' or track['dataType'] == 'gtf' or track['dataType'] == 'blastxml': - p = subprocess.Popen(['flatfile-to-json.pl', '--gff', gff3_file, '--trackType', metadata['type'], '--trackLabel', metadata['label'], '--Config', '{"glyph": "JBrowse/View/FeatureGlyph/Segments", "category" : "%s"}' % metadata['category'], '--clientConfig', '{"color" : "%s"}' % metadata['color'], '--out', self.json]) + flat_file = os.path.join(self.raw, track['fileName']) + if track['dataType'] == 'bed': + subprocess.call(['flatfile-to-json.pl', '--bed', flat_file, '--trackType', metadata['type'], '--trackLabel', metadata['label'], '--Config', '{"category" : "%s"}' % metadata['category'], '--clientConfig', '{"color" : "%s"}' % metadata['color'], '--out', self.json]) + elif track['dataType'] == 'bedSpliceJunctions' or track['dataType'] == 'gtf' or track['dataType'] == 'blastxml': + subprocess.call(['flatfile-to-json.pl', '--gff', flat_file, '--trackType', metadata['type'], '--trackLabel', metadata['label'], '--Config', '{"glyph": "JBrowse/View/FeatureGlyph/Segments", "category" : "%s"}' % metadata['category'], '--clientConfig', '{"color" : "%s"}' % metadata['color'], '--out', self.json]) elif track['dataType'] == 'gff3_transcript': - p = subprocess.Popen(['flatfile-to-json.pl', '--gff', gff3_file, '--trackType', metadata['type'], '--trackLabel', metadata['label'], '--Config', '{"transcriptType": "transcript", "category" : "%s"}' % metadata['category'], '--clientConfig', '{"color" : "%s"}' % metadata['color'], '--out', self.json]) + subprocess.call(['flatfile-to-json.pl', '--gff', flat_file, '--trackType', metadata['type'], '--trackLabel', metadata['label'], '--Config', '{"transcriptType": "transcript", "category" : "%s"}' % metadata['category'], '--clientConfig', '{"color" : "%s"}' % metadata['color'], '--out', self.json]) else: - p = subprocess.Popen(['flatfile-to-json.pl', '--gff', gff3_file, '--trackType', metadata['type'], '--trackLabel', metadata['label'], '--Config', '{"category" : "%s"}' % metadata['category'], '--clientConfig', '{"color" : "%s"}' % metadata['color'], '--out', self.json]) - p.communicate() + subprocess.call(['flatfile-to-json.pl', '--gff', flat_file, '--trackType', metadata['type'], '--trackLabel', metadata['label'], '--Config', '{"category" : "%s"}' % metadata['category'], '--clientConfig', '{"color" : "%s"}' % metadata['color'], '--out', self.json]) def indexName(self): - p = subprocess.Popen(['generate-names.pl', '-v', '--out', self.json]) - p.communicate() + subprocess.call(['generate-names.pl', '-v', '--out', self.json]) print "finished name index \n" def makeArchive(self): @@ -83,8 +82,7 @@ except OSError as oserror: print "Cannot create data folder({0}): {1}".format(oserror.errno, oserror.strerror) shutil.copytree(self.out_path, data_folder) - p = subprocess.Popen(['chmod', '-R', 'o+rx', '/var/www/html/JBrowse-1.12.1/jbrowse_hub']) - p.communicate() + subprocess.call(['chmod', '-R', 'o+rx', '/var/www/html/JBrowse-1.12.1/jbrowse_hub']) shutil.rmtree(self.out_path) #TODO: this will list all zip files in the filedir and sub-dirs. worked in Galaxy but all list zip files in test-data when @@ -93,7 +91,7 @@ with open(self.outfile, 'w') as htmlfile: htmlstr = 'The JBrowse Hub is created: <br>' zipfiles = '<li><a href = "%s">Download</a></li>' - jbrowse_hub = '<li><a href = "http://192.168.56.11/JBrowse-1.12.1/index.html?data=jbrowse_hub/json" target="_blank">View JBrowse Hub</a></li>' + jbrowse_hub = '<li><a href = "/static/JBrowse-1.12.1/index.html?data=jbrowse_hub/json" target="_blank">View JBrowse Hub</a></li>' filedir_abs = os.path.abspath(self.outfile) filedir = os.path.dirname(filedir_abs) filedir = os.path.join(filedir, self.outfolder) @@ -122,8 +120,11 @@ bam_track['baiUrlTemplate'] = os.path.join('../raw', track['index']) bam_track['label'] = metadata['label'] bam_track['category'] = metadata['category'] - utils.add_tracks_to_json(json_file, bam_track, 'add_tracks') - + bam_track = json.dumps(bam_track) + #Use add-track-json.pl to add bam track to json file + new_track = subprocess.Popen(['echo', bam_track], stdout=subprocess.PIPE) + subprocess.call(['add-track-json.pl', json_file], stdin=new_track.stdout) + def BigWig(self, track, metadata): #create trackList.json if not exist self.createTrackList() @@ -135,7 +136,11 @@ bigwig_track['label'] = metadata['label'] bigwig_track['style'] = metadata['style'] bigwig_track['category'] = metadata['category'] - utils.add_tracks_to_json(json_file, bigwig_track, 'add_tracks') + bigwig_track = json.dumps(bigwig_track) + #Use add-track-json.pl to add bigwig track to json file + new_track = subprocess.Popen(['echo', bigwig_track], stdout=subprocess.PIPE) + #output = new_track.communicate()[0] + subprocess.call(['add-track-json.pl', json_file], stdin=new_track.stdout) #If the metadata is not set, use the default value def SetMetadata(self, track, metadata): @@ -147,9 +152,9 @@ if 'style' not in metadata.keys(): metadata['style'] = {} if 'pos_color' not in metadata['style'] or metadata['style']['pos_color'] == '': - metadata['pos_color'] = "#FFA600" + metadata['style']['pos_color'] = "#FFA600" if 'neg_color' not in metadata['style'] or metadata['style']['neg_color'] == '': - metadata['neg_color'] = "#005EFF" + metadata['style']['neg_color'] = "#005EFF" if 'category' not in metadata.keys() or metadata['category'] == '': metadata['category'] = "Default group" if track['dataType'] == 'blastxml':
--- a/jbrowse_hub.py Wed Mar 29 14:46:23 2017 -0400 +++ b/jbrowse_hub.py Thu Mar 30 16:03:16 2017 -0400 @@ -27,11 +27,17 @@ #Tool Directory parser.add_argument('-d', '--tool_directory', help='The directory of JBrowse file convertion scripts and UCSC tools') + #GFF3 + parser.add_argument('--gff3', action='append', help='GFF3 format') + # GFF3 structure: gene->transcription->CDS - parser.add_argument('--gff3_transcript', action='append', help='GFF3 format, structure: gene->transcription->CDS') + parser.add_argument('--gff3_transcript', action='append', help='GFF3 format for gene prediction, structure: gene->transcription->CDS') # GFF3 structure: gene->mRNA->CDS - parser.add_argument('--gff3_mrna', action='append', help='GFF3 format, structure: gene->mRNA->CDS') + parser.add_argument('--gff3_mrna', action='append', help='GFF3 format for gene prediction, structure: gene->mRNA->CDS') + + # generic BED + parser.add_argument('--bed', action='append', help='BED format') # trfBig simple repeats (BED 4+12) parser.add_argument('--bedSimpleRepeats', action='append', help='BED 4+12 format, using simpleRepeats.as') @@ -93,9 +99,11 @@ all_tracks = trackObject.trackObject(chrom_size.name, genome, extra_files_path) array_inputs_bam = args.bam + array_inputs_bed = args.bed array_inputs_bed_simple_repeats = args.bedSimpleRepeats array_inputs_bed_splice_junctions = args.bedSpliceJunctions array_inputs_bigwig = args.bigwig + array_inputs_gff3 = args.gff3 array_inputs_gff3_transcript = args.gff3_transcript array_inputs_gff3_mrna = args.gff3_mrna array_inputs_gtf = args.gtf @@ -103,12 +111,16 @@ if array_inputs_bam: all_datatype_dictionary['bam'] = array_inputs_bam + if array_inputs_bed: + all_datatype_dictionary['bed'] = array_inputs_bed if array_inputs_bed_simple_repeats: all_datatype_dictionary['bedSimpleRepeats'] = array_inputs_bed_simple_repeats if array_inputs_bed_splice_junctions: all_datatype_dictionary['bedSpliceJunctions'] = array_inputs_bed_splice_junctions if array_inputs_bigwig: all_datatype_dictionary['bigwig'] = array_inputs_bigwig + if array_inputs_gff3: + all_datatype_dictionary['gff3'] = array_inputs_gff3 if array_inputs_gff3_transcript: all_datatype_dictionary['gff3_transcript'] = array_inputs_gff3_transcript if array_inputs_gff3_mrna:
--- a/jbrowse_hub.xml Wed Mar 29 14:46:23 2017 -0400 +++ b/jbrowse_hub.xml Thu Mar 30 16:03:16 2017 -0400 @@ -42,8 +42,11 @@ #set extra_data_dict = {"label" : $track_label, "category" : $group_name} #if $f.formatChoice.format_select == 'bed' #set track_color = str($f.formatChoice.track_color) + #if $f.formatChoice.bedChoice.bed_select == 'bed_generic_option' + --bed $f.formatChoice.bedChoice.BED_generic + #silent $prepare_json($f.formatChoice.bedChoice.BED_generic, extra_data_dict) #silent extra_data_dict.update({"color" : $track_color}) - #if $f.formatChoice.bedChoice.bed_select == 'bed_simple_repeats_option' + #elif $f.formatChoice.bedChoice.bed_select == 'bed_simple_repeats_option' --bedSimpleRepeats $f.formatChoice.bedChoice.BED_simple_repeats #silent $prepare_json($f.formatChoice.bedChoice.BED_simple_repeats, extra_data_dict) #elif $f.formatChoice.bedChoice.bed_select == 'bed_splice_junctions_option' @@ -58,7 +61,10 @@ #if $f.formatChoice.format_select == 'gff3' #set track_color = str($f.formatChoice.track_color) #silent extra_data_dict.update({"color" : $track_color}) - #if $f.formatChoice.gff3Choice.gff3_select == 'gff3_transcript' + #if $f.formatChoice.gff3Choice.gff3_select == 'gff3_generic' + --gff3 $f.formatChoice.gff3Choice.GFF3_generic + #silent $prepare_json($f.formatChoice.gff3Choice.GFF3_generic, extra_data_dict) + #esif $f.formatChoice.gff3Choice.gff3_select == 'gff3_transcript' --gff3_transcript $f.formatChoice.gff3Choice.GFF3_transcript #silent $prepare_json($f.formatChoice.gff3Choice.GFF3_transcript, extra_data_dict) #elif $f.formatChoice.gff3Choice.gff3_select == 'gff3_mrna' @@ -121,9 +127,18 @@ <when value="bed"> <conditional name="bedChoice"> <param name="bed_select" type="select" label="Bed Choice"> + <option value="bed_generic_option">BED format</option> <option value="bed_simple_repeats_option">BED Simple repeat (bed4+12 / simpleRepeat.as)</option> <option value="bed_splice_junctions_option">BED Splice junctions (bed12+1 / spliceJunctions.as)</option> </param> + <when value="bed_generic_option"> + <param + format="bed" + name="BED_generic" + type="data" + label="Bed File" + /> + </when> <when value="bed_simple_repeats_option"> <param format="bed" @@ -185,15 +200,24 @@ <when value="gff3"> <conditional name="gff3Choice"> <param name="gff3_select" type="select" label="gff3 type"> - <option value="gff3_transcript">GFF3 format, structure: gene->transcription->CDS</option> - <option value="gff3_mrna">GFF3 format, structure: gene->mRNA->CDS</option> + <option value="gff3_generic">GFF3 format</option> + <option value="gff3_transcript">GFF3 format output from gene prediction tools (e.g. Augustus), structure: gene->transcription->CDS</option> + <option value="gff3_mrna">GFF3 format output from gene prediction tools (e.g. SNAP), structure: gene->mRNA->CDS</option> </param> + <when value="gff3_generic"> + <param + format="gff3" + name="GFF3_generic" + type="data" + label="GFF3 File" + /> + </when> <when value="gff3_transcript"> <param format="gff3" name="GFF3_transcript" type="data" - label="GFF3 File" + label="GFF3 File from gene prediction" /> </when> <when value="gff3_mrna"> @@ -201,7 +225,7 @@ format="gff3" name="GFF3_mrna" type="data" - label="GFF3 File" + label="GFF3 File from gene prediction" /> </when> </conditional>
--- a/trackObject.py Wed Mar 29 14:46:23 2017 -0400 +++ b/trackObject.py Thu Mar 30 16:03:16 2017 -0400 @@ -33,10 +33,12 @@ fileName = os.path.basename(dataFile) des_path = os.path.join(self.raw_folder, fileName) track = {} - if dataType == 'gff3_mrna' or dataType == 'gff3_transcript' or dataType == 'fasta' or dataType == 'bam' or dataType == 'bigwig' or dataType == 'bai': + if dataType == 'bed' or dataType == 'gff3' or dataType == 'gff3_mrna' or dataType == 'gff3_transcript' or dataType == 'fasta' or dataType == 'bam' or dataType == 'bigwig': if dataType == 'bam': # JBrowse will raise error: not a BAM file if the filename hasn't .bam extension - fileName = fileName + '.bam' + extension = os.path.splitext(fileName)[1] + if extension != '.bam': + fileName = fileName + '.bam' des_path = os.path.join(self.raw_folder, fileName) bam_index = utils.createBamIndex(dataFile) indexname = os.path.basename(bam_index)
--- a/utils.py Wed Mar 29 14:46:23 2017 -0400 +++ b/utils.py Thu Mar 30 16:03:16 2017 -0400 @@ -35,13 +35,11 @@ except IOError as err: print "Cannot create tempfile err({0}): {1}".format(err.errno, err.strerror) try: - p = subprocess.Popen(['faToTwoBit', reference, twoBitFile.name]) - p.communicate() + subprocess.call(['faToTwoBit', reference, twoBitFile.name]) except OSError as err: print "Cannot generate twoBitFile from faToTwoBit err({0}): {1}".format(err.errno, err.strerror) try: - p = subprocess.Popen(['twoBitInfo', twoBitFile.name, chrom_sizes.name]) - p.communicate() + subprocess.call(['twoBitInfo', twoBitFile.name, chrom_sizes.name]) except OSError as err: print "Cannot generate chrom_sizes from twoBitInfo err({0}): {1}".format(err.errno, err.strerror) return chrom_sizes @@ -155,8 +153,7 @@ return "gonramp_" + sanitized_name def createBamIndex(bamfile): - p = subprocess.Popen(['samtools', 'index', bamfile]) - p.communicate() + subprocess.call(['samtools', 'index', bamfile]) filename = bamfile + '.bai' if os.path.exists(filename): return filename