def testParseHeaderLinesNoTracks(self):
self.assertEquals(
('track type', 'segments'),
GSuiteParser._parseHeaderLine('##track type: segments\n'))
self.assertEquals(('location', 'local'),
GSuiteParser._parseHeaderLine('##location:Local\n'))
self.assertEquals(
('file format', 'primary'),
GSuiteParser._parseHeaderLine('##File format: primary\n'))
self.assertEquals(('genome', 'HG18'),
GSuiteParser._parseHeaderLine('##genome: HG18\n'))
def _validateGSuiteFile(galaxyTN):
import gold.gsuite.GSuiteParser as GSuiteParser
from quick.application.ExternalTrackManager import ExternalTrackManager
from cStringIO import StringIO
galaxyFn = ExternalTrackManager.extractFnFromGalaxyTN(galaxyTN)
outFile = StringIO()
ok = GSuiteParser.validate(galaxyFn,
outFile=outFile,
printHelpText=False)
if not ok:
return outFile.getvalue()
def testParseAndCompose(self):
inputContents = \
'##location: multiple\n' \
'##file format: multiple\n' \
'##track type: unknown\n' \
'##genome: hg18\n' \
'###uri\ttitle\tfile_format\tcell\tantibody\textra\n' \
'ftp://server.somewhere.com/path/to/file1.bed\tTrack1\tprimary\tk562\tcMyb\t.\n' \
'http://server.other.com/path/to/file2.bed\tTrack2\tprimary\tGM12878\tcMyc\t.\n' \
'https://server.other.com/path/to/file3.bed\tfile3.bed\tprimary\tGM12878\tcMyb\t.\n' \
'rsync://server.other.com/path/to/file4;wig\tTrack4\tprimary\tNHFL\t.\t.\n' \
'hb:/track/name/hierarchy\tTrack5\tpreprocessed\t.\t.\t.\n' \
'galaxy:/ad123dd12fg;btrack?track=track%3Aname\tTrack6\tpreprocessed\tk562\tcMyb\t.\n' \
'file:/path/to/file.btrack?track=track%3Aname\tTrack name7\tpreprocessed\t.\tcMyb\tyes\n'
gSuite = GSuiteParser.parseLines(inputContents.split('\n'))
outputContents = GSuiteComposer.composeToString(gSuite)
self.assertEquals(inputContents, outputContents)
def testParseColSpecLineDirectlyMoreStdCols(self):
colSpecLine = '###Uri\tTitle\tfile_format\ttrack_type\tgenome\tANTIBODY'
self.assertEquals([
'uri', 'title', 'file_format', 'track_type', 'genome', 'antibody'
], GSuiteParser._parseColumnSpecLine(colSpecLine))
def testParseColSpecLineDirectly(self):
colSpecLine = '###uri\tantibody'
self.assertEquals(['uri', 'antibody'],
GSuiteParser._parseColumnSpecLine(colSpecLine))
def testParseTrackLineFull(self):
contents = \
'###uri\ttitle\tcell\tantibody\n' \
'ftp://server.somewhere.com/path/to/file1.bed.gz\tTrack\tk562\tcMyb\n' \
'http://server.other.com/path/to/file2.bed?query=something\tTrack2\tGM12878\tcMyc\n' \
'https://server.other.com/path/to/file3.bed?query=something\tTrack3\tGM12878\tcMyb\n' \
'rsync://server.other.com/path/to/file4;wig\tTrack4\tNHFL\t.\n' \
'hb:/track/name/hierarchy\tTrack (2)\t.\t.\n' \
'galaxy:/ad123dd12fg;btrack?track=track:name\tTrack (3)\tk562\tcMyb\n' \
'file:/path/to/file.btrack?track=track:name\tTrack name7\t.\tcMyb\n'
gSuite = self._parseContents(contents)
tracks = list(gSuite.allTracks())
self.assertEquals(7, len(tracks))
self._commonAssertTrack(
tracks[0],
uri='ftp://server.somewhere.com/path/to/file1.bed.gz',
scheme='ftp',
netloc='server.somewhere.com',
path='/path/to/file1.bed.gz',
query=None,
suffix='gz',
trackName=None,
title='Track',
location='remote',
fileFormat='primary',
trackType='unknown',
genome='unknown',
attributes=OrderedDict([('cell', 'k562'), ('antibody', 'cMyb')]))
self._commonAssertTrack(
tracks[1],
uri='http://server.other.com/path/to/file2.bed?query=something',
scheme='http',
netloc='server.other.com',
path='/path/to/file2.bed',
query='query=something',
suffix='bed',
trackName=None,
title='Track2',
location='remote',
fileFormat='primary',
trackType='unknown',
genome='unknown',
attributes=OrderedDict([('cell', 'GM12878'),
('antibody', 'cMyc')]))
self._commonAssertTrack(
tracks[2],
uri='https://server.other.com/path/to/file3.bed?query=something',
scheme='https',
netloc='server.other.com',
path='/path/to/file3.bed',
query='query=something',
suffix='bed',
trackName=None,
title='Track3',
location='remote',
fileFormat='primary',
trackType='unknown',
genome='unknown',
attributes=OrderedDict([('cell', 'GM12878'),
('antibody', 'cMyb')]))
self._commonAssertTrack(
tracks[3],
uri='rsync://server.other.com/path/to/file4;wig',
scheme='rsync',
netloc='server.other.com',
path='/path/to/file4',
query=None,
suffix='wig',
trackName=None,
title='Track4',
location='remote',
fileFormat='primary',
trackType='unknown',
genome='unknown',
attributes=OrderedDict([('cell', 'NHFL')]))
self._commonAssertTrack(tracks[4],
uri='hb:/track/name/hierarchy',
scheme='hb',
netloc=None,
path=None,
query=None,
suffix=None,
trackName=['track', 'name', 'hierarchy'],
title='Track (2)',
location='local',
fileFormat='preprocessed',
trackType='unknown',
genome='unknown',
attributes=OrderedDict())
self._commonAssertTrack(
tracks[5],
uri='galaxy:/ad123dd12fg;btrack?track=track%3Aname',
scheme='galaxy',
netloc=None,
path='/path/to/dataset_ad123dd12fg.dat',
query='track=track:name',
suffix='btrack',
trackName=['track', 'name'],
title='Track (3)',
location='local',
fileFormat='preprocessed',
trackType='unknown',
genome='unknown',
attributes=OrderedDict([('cell', 'k562'), ('antibody', 'cMyb')]))
self._commonAssertTrack(
tracks[6],
uri='file:/path/to/file.btrack?track=track%3Aname',
scheme='file',
netloc=None,
path='/path/to/file.btrack',
query='track=track:name',
suffix='btrack',
trackName=['track', 'name'],
title='Track name7',
location='local',
fileFormat='preprocessed',
trackType='unknown',
genome='unknown',
attributes=OrderedDict([('antibody', 'cMyb')]))
self.assertEquals('multiple', gSuite.location)
self.assertEquals('multiple', gSuite.fileFormat)
self.assertEquals('unknown', gSuite.trackType)
self.assertEquals('unknown', gSuite.genome)
self.assertEquals(['cell', 'antibody'], gSuite.attributes)
from cStringIO import StringIO
GSuiteParser.validateFromString(contents, outFile=StringIO())
def _parseContents(contents):
return GSuiteParser.parseLines(contents.split('\n'))
def getGSuiteFromGSuiteFile(gSuiteFn):
from gold.gsuite import GSuiteParser
return GSuiteParser.parse(gSuiteFn)
attributes=OrderedDict([('a', 'yes'), ('b', 'no')])))
uri2 = HbGSuiteTrack.generateURI(
trackName=['Genes and gene subsets', 'Genes', 'Refseq'])
gSuite.addTrack(
GSuiteTrack(uri2, attributes=OrderedDict([('b', 'no'), ('c', 'yes')])))
gSuite.setGenomeOfAllTracks('hg19')
contents = GSuiteComposer.composeToString(gSuite)
print 'GSuite file contents'
print '--------------------'
print contents
gSuite2 = GSuiteParser.parseFromString(contents)
print 'Various ways of direct access'
print '-----------------------------'
print "genome=%s, location=%s, file format=%s, track type=%s, attributes=%s" % \
(gSuite.genome, gSuite.location, gSuite.fileFormat, gSuite.trackType, gSuite.attributes)
for track in gSuite2.allTracks():
print "uri=%s, path=%s, trackName=%s" % (track.uri, track.path,
track.trackName)
print "netloc=" + gSuite2.getTrackFromTitle('Track1').netloc
tracks = list(gSuite.allTracks())
print "b=" + tracks[1].attributes['b']