def _constructBins(regSpec, binSpec, genome, trackNames):
# Construct and check bins
try:
from quick.application.GalaxyInterface import GalaxyInterface
userBinSource = GalaxyInterface._getUserBinSource(regSpec, binSpec, genome, trackNames)
return [None, userBinSource]
except Exception, e:
results = Results([], [], '')
results.addError(InvalidRunSpecException('Error in specification of analysis region or binsize: ' + str(e)))
logMessage('Error in specification of analysis region (' + regSpec +') or binsize: (' + binSpec + ')')
if DebugConfig.PASS_ON_BATCH_EXCEPTIONS:
raise
return [results, None]
def _constructBins(regSpec, binSpec, genome, trackName1, trackName2):
#Construct and check bins
try:
#userBinSource= UserBinSource(regSpec, binSpec)
from quick.application.GalaxyInterface import GalaxyInterface
# from config.Config import DEFAULT_GENOME
userBinSource = GalaxyInterface._getUserBinSource(regSpec, binSpec, genome, trackName1, trackName2)
return [None, userBinSource]
except Exception, e:
#results = Results(trackName1, trackName2, statClassName)
results = Results([],[],'')
results.addError(InvalidRunSpecException('Error in specification of analysis region or binsize: ' + str(e)))
logMessage('Error in specification of analysis region (' + regSpec +') or binsize: (' + binSpec + ')')
if DebugConfig.PASS_ON_BATCH_EXCEPTIONS:
raise
return [results, None]
def runJob(batchLine, genome, fullAccess, galaxyFn=None, printProgress=True):
bc = BatchRunner.parseBatchLine(batchLine, genome, fullAccess)
if bc.errorResult is not None:
return bc.errorResult
#Try a full run, and return either results or an exception
try:
#track = Track(trackName1)
#track2 = Track(trackName2)
#if 'tf1' in paramDict:
# track.setFormatConverter(formatConverter)
#results = StatRunner.run(userBinSource , Track(trackName1), Track(trackName2), \
# wrapClass(STAT_CLASS_DICT[statClassName], keywords=paramDict) )
#results = StatRunner.run(userBinSource , track, track2, \
# wrapClass(STAT_CLASS_DICT[statClassName], keywords=paramDict) )
fullRunParams = {}
if USE_PARALLEL:
#if galaxyFn == None: #then this is a test
uniqueId = time.time()
#else:
#uniqueId = extractIdFromGalaxyFn(galaxyFn)[1]
fullRunParams["uniqueId"] = uniqueId
if bc.cleanedTrackNameIntensity is not None:
fullRunParams['trackNameIntensity'] = '|'.join(tuple(bc.cleanedTrackNameIntensity))
analysisDefParams = [ '[' + key + '=' + value + ']' for key,value in bc.paramDict.items()]
analysisDef = ''.join(analysisDefParams) + '->' + bc.statClassName
from quick.application.GalaxyInterface import GalaxyInterface
GalaxyInterface._tempAnalysisDefHacks(analysisDef)
if printProgress:
print 'Corresponding batch command line:<br>' + \
GalaxyInterface._revEngBatchLine(bc.trackName1, bc.trackName2, bc.trackNameIntensity, analysisDef, bc.regSpec, bc.binSpec, genome) + '<br><br>'
results = AnalysisDefJob(analysisDef, bc.cleanedTrackName1, bc.cleanedTrackName2, bc.userBinSource, galaxyFn=galaxyFn, **fullRunParams).run(printProgress)
presCollectionType = results.getPresCollectionType()
if len(results.getResDictKeys()) > 0 and GalaxyInterface.APPEND_ASSEMBLY_GAPS and presCollectionType=='standard':
if USE_PARALLEL:
gapRes = AssemblyGapJob(bc.userBinSource, genome, uniqueId=uniqueId).run(printProgress)
else:
gapRes = AssemblyGapJob(bc.userBinSource, genome).run(printProgress)
results.includeAdditionalResults(gapRes, ensureAnalysisConsistency=False)
except Exception, e:
#print 'NOWAG BExc'
results = Results(bc.cleanedTrackName1, bc.cleanedTrackName2, bc.statClassName)
results.addError(e)
logException(e,message='Error in batch run')
if DebugConfig.PASS_ON_BATCH_EXCEPTIONS:
raise
return results
def parseBatchLine(batchLine, genome, fullAccess):
if batchLine[0] == '#' or batchLine.strip() == '':
return
from urllib import unquote
#Split and check number of columns
cols = [x for x in batchLine.strip().split(BATCH_COL_SEPARATOR)]
if len(cols) != 6:
results = Results(['N/A'], ['N/A'], 'N/A')
#results.addResultComponent( 'Invalid',InvalidRunResultComponent('Error in batch specification. 6 columns are required, while '\
# + str(len(cols)) + ' are given.'))
results.addError(InvalidRunSpecException('Error in batch specification. 6 columns are required, while '\
+ str(len(cols)) + ' are given: ' + batchLine))
return results, None, None, None, None
bc = BatchContents()
bc.regSpec = cols[1]
bc.binSpec = unquote(cols[2])
from quick.application.ExternalTrackManager import ExternalTrackManager
if ExternalTrackManager.isGalaxyTrack(bc.binSpec.split(':')):
bc.binSpec = ExternalTrackManager.extractFnFromGalaxyTN(
bc.binSpec.split(':'))
try:
from quick.application.GalaxyInterface import GalaxyInterface
bc.trackName1 = [unquote(x) for x in cols[3].split(':')]
bc.trackName2 = [unquote(x) for x in cols[4].split(':')]
bc.cleanedTrackName1, bc.cleanedTrackName2 = GalaxyInterface._cleanUpTracks(
[bc.trackName1, bc.trackName2], genome, realPreProc=True)
bc.cleanedTrackName1 = BatchRunner._inferTrackName(
bc.cleanedTrackName1, genome, fullAccess)
bc.cleanedTrackName2 = BatchRunner._inferTrackName(
bc.cleanedTrackName2, genome, fullAccess)
except (InvalidRunSpecException, IdenticalTrackNamesError), e:
if DebugConfig.PASS_ON_BATCH_EXCEPTIONS:
raise
bc.errorResult = Results(['N/A'], ['N/A'], 'N/A')
bc.errorResult.addError(e)
return bc
def parseBatchLine(batchLine, genome, fullAccess):
if batchLine[0] == '#' or batchLine.strip() == '':
return
from urllib import unquote
# Split and check number of columns
cols = [x for x in batchLine.strip().split(BATCH_COL_SEPARATOR)]
if len(cols) != 6:
results = Results(['N/A'], ['N/A'], 'N/A')
results.addError(InvalidRunSpecException('Error in batch specification. 6 columns are required, while '\
+ str(len(cols)) + ' are given: ' + batchLine))
return results, None, None, None, None
bc = BatchContents()
bc.regSpec = cols[1]
bc.binSpec = unquote(cols[2])
from quick.application.ExternalTrackManager import ExternalTrackManager
if ExternalTrackManager.isGalaxyTrack(bc.binSpec.split(':')):
bc.binSpec = ExternalTrackManager.extractFnFromGalaxyTN(bc.binSpec.split(':'))
bc.statClassName, bc.paramDict = BatchRunner._parseClassAndParams(cols[5])
bc.trackNames = [[unquote(x) for x in cols[i].split(':')] for i in [3, 4]]
if 'trackNameIntensity' in bc.paramDict:
bc.trackNames.append(convertTNstrToTNListFormat(bc.paramDict['trackNameIntensity'], doUnquoting=True))
from quick.application.GalaxyInterface import GalaxyInterface
partlyCleanedTrackNames = GalaxyInterface._cleanUpTracks(bc.trackNames, genome, realPreProc=True)
try:
bc.cleanedTrackNames = BatchRunner._inferTrackNames(partlyCleanedTrackNames, genome, fullAccess)
except (InvalidRunSpecException,IdenticalTrackNamesError), e:
if DebugConfig.PASS_ON_BATCH_EXCEPTIONS:
raise
bc.errorResult = Results(['N/A'],['N/A'],'N/A')
bc.errorResult.addError(e)
return bc
def runJob(batchLine, genome, fullAccess):
if batchLine[0] == '#' or batchLine.strip()=='':
return
from urllib import unquote
#Split and check number of columns
cols = [x for x in batchLine.strip().split(BATCH_COL_SEPARATOR)]
if len(cols) != 6:
results = Results(['N/A'],['N/A'],'N/A')
#results.addResultComponent( 'Invalid',InvalidRunResultComponent('Error in batch specification. 6 columns are required, while '\
# + str(len(cols)) + ' are given.'))
results.addError(InvalidRunSpecException('Error in batch specification. 6 columns are required, while '\
+ str(len(cols)) + ' are given: ' + batchLine))
return results
#print 'COL2: ',cols[2]
cols[2] = unquote(cols[2])
#print 'COL2: ',cols[2]
from quick.application.ExternalTrackManager import ExternalTrackManager
if ExternalTrackManager.isGalaxyTrack(cols[2].split(':')):
cols[2] = ExternalTrackManager.extractFnFromGalaxyTN(cols[2].split(':'))
#print 'COL2: ',cols[2]
try:
from quick.application.GalaxyInterface import GalaxyInterface
trackName1 = [unquote(x) for x in cols[3].split(':')]
trackName2 = [unquote(x) for x in cols[4].split(':')]
cleanedTrackName1, cleanedTrackName2 = GalaxyInterface._cleanUpTracks([trackName1, trackName2], genome, realPreProc=True)
cleanedTrackName1 = BatchRunner._inferTrackName(':'.join(cleanedTrackName1), genome, fullAccess)
cleanedTrackName2 = BatchRunner._inferTrackName(':'.join(cleanedTrackName2), genome, fullAccess)
except (InvalidRunSpecException,IdenticalTrackNamesError), e:
if DebugConfig.PASS_ON_BATCH_EXCEPTIONS:
raise
results = Results(['N/A'],['N/A'],'N/A')
results.addError(e)
return results
def runJob(batchLine, genome, fullAccess, galaxyFn=None, printProgress=True):
bc = BatchRunner.parseBatchLine(batchLine, genome, fullAccess)
if bc.errorResult is not None:
return bc.errorResult
#Try a full run, and return either results or an exception
try:
#track = Track(trackName1)
#track2 = Track(trackName2)
#if 'tf1' in paramDict:
# track.setFormatConverter(formatConverter)
#results = StatRunner.run(userBinSource , Track(trackName1), Track(trackName2), \
# wrapClass(STAT_CLASS_DICT[statClassName], keywords=paramDict) )
#results = StatRunner.run(userBinSource , track, track2, \
# wrapClass(STAT_CLASS_DICT[statClassName], keywords=paramDict) )
fullRunParams = {}
if USE_PARALLEL:
# TODO: Requirements for parallel runs should not be added in places like these. Parallelization
# should be a feature of the job runner somehow
#if galaxyFn == None: #then this is a test
uniqueId = time.time()
#else:
#uniqueId = extractIdFromGalaxyFn(galaxyFn)[1]
fullRunParams["uniqueId"] = uniqueId
from quick.application.GalaxyInterface import GalaxyInterface
analysisDefParams = [ '[' + key + '=' + value + ']' for key,value in bc.paramDict.items()]
analysisDef = ''.join(analysisDefParams) + '->' + bc.statClassName
# TODO: Keeping the ugly accesses to private methods in GalaxyInterface for now. To be refactored.
trackNames, analysisDef = GalaxyInterface._cleanUpAnalysisDef(bc.cleanedTrackNames, analysisDef)
if printProgress:
revEngBatchLine = RunDescription.getRevEngBatchLine(
analysisDef, bc.trackNames, bc.cleanedTrackNames, bc.regSpec, bc.binSpec, genome
)
print 'Corresponding batch command line:<br>{}<br><br>'.format(revEngBatchLine)
results = AnalysisDefJob(analysisDef, bc.cleanedTrackNames[0], bc.cleanedTrackNames[1], bc.userBinSource, galaxyFn=galaxyFn, **fullRunParams).run(printProgress)
presCollectionType = results.getPresCollectionType()
if len(results.getResDictKeys()) > 0 and GalaxyInterface.APPEND_ASSEMBLY_GAPS and presCollectionType=='standard':
if USE_PARALLEL:
gapRes = AssemblyGapJob(bc.userBinSource, genome, uniqueId=uniqueId).run(printProgress)
else:
gapRes = AssemblyGapJob(bc.userBinSource, genome).run(printProgress)
results.includeAdditionalResults(gapRes, ensureAnalysisConsistency=False)
except Exception, e:
#print 'NOWAG BExc'
results = Results(bc.cleanedTrackNames[0], bc.cleanedTrackNames[1], bc.statClassName)
results.addError(e)
logException(e,message='Error in batch run')
if DebugConfig.PASS_ON_BATCH_EXCEPTIONS:
raise
return results
def executeSelfFeature(cls, genome, tracks, track_names, clusterMethod,
extra_option, feature, distanceType, kmeans_alg,
galaxyFn, regSpec, binSpec):
from proto.RSetup import r
#regSpec, binSpec = 'bed', '/usit/invitro/data/galaxy/galaxy-dist-hg-dev/./database/files/017/dataset_17084.dat'
silenceRWarnings()
jobFile = open(galaxyFn, 'w')
# print>>jobFile, 'PARAMS: ', dict(zip('genome, tracks, track_names, clusterMethod, extra_option, feature, distanceType, kmeans_alg, regSpec, binSpec'.split(','), [repr(v)+'<br>'for v in [genome, tracks, track_names, clusterMethod, extra_option, feature, distanceType, kmeans_alg,regSpec, binSpec]])), '<br><br>'
batchRun = GalaxyRunSpecificFile(['batch_run_job.txt'], galaxyFn)
print >> jobFile, '<h3>Results for the "similarity of positional distribution along the genome" way of clustering<h3/><br/><br/>'
with open(batchRun.getDiskPath(ensurePath=True), 'w') as batchFile:
print >> batchFile, '$clusterBySelfFeature', (genome, '$'.join([
':'.join(t) for t in tracks
]), ':'.join(track_names), clusterMethod, extra_option, feature,
distanceType,
kmeans_alg, regSpec,
binSpec)
print >> jobFile, batchRun.getLink(
'View batch script line for this analysis<br/>')
#print>>jobFile, 'Batch script syntax for this analysis:<br>$clusterBySelfFeature', (genome, '$'.join([':'.join(t) for t in tracks]), ':'.join(track_names) , clusterMethod, extra_option, feature, distanceType, kmeans_alg, regSpec, binSpec), '<br><br>'
#print>>jobFile, 'signature of method clusterBySelfFeature:<br>', 'clusterBySelfFeature(genome, tracksStr, track_namesStr, clusterMethod, extra_option, feature, distanceType, kmeans_alg, regSpec, binSpec):<br><br><br>'
prettyTrackNames = [
v[-1].replace('RoadMap_', '').replace('.H3K4me1', '')
for v in tracks
]
#prettyTrackNames = [prettyPrintTrackName(v, shortVersion=True) for v in tracks]
f_matrix = cls.construct_feature_matrix(genome, tracks, feature,
regSpec, binSpec)
#print>>jobFile, 'dir f_matrix: ', dir(f_matrix), regSpec, binSpec
userBinSource = GalaxyInterface._getUserBinSource(
regSpec, binSpec, genome)
binNames = [
str(bin)
for binIndex, bin in enumerate(sorted(list(userBinSource)))
]
if len(binNames) != f_matrix.shape[1]:
binNames = ['Microbin' + str(i) for i in range(f_matrix.shape[1])]
r.assign('bin_names', binNames)
r.assign('track_names', prettyTrackNames
) #use as track names, will be shown in clustering figure
r.assign('f_matrix', f_matrix)
r.assign('distanceType', distanceType)
r('row.names(f_matrix) <- track_names')
r('colnames(f_matrix) <- bin_names')
if clusterMethod == 'Hierarchical clustering' and extra_option != "--select--":
#print 'galaxyFn: ', galaxyFn
figure = GalaxyRunSpecificFile(
['cluster_tracks_result_figure.pdf'], galaxyFn)
figurepath = figure.getDiskPath(ensurePath=True)
r('d <- dist(f_matrix, method=distanceType)')
distTable = r('d')
distMatrix = GalaxyRunSpecificFile(['distance_matrix_result.txt'],
galaxyFn)
distMatrixPath = distMatrix.getDiskPath(True)
open(distMatrixPath, 'w').write(str(distTable))
print >> jobFile, distMatrix.getLink(
'View the distance matrix for this analysis <br>')
#with open(distMatrixPath,'w') as distObj:
# #distTable = d_matrix.tolist()
# core = HtmlCore()
# core.tableHeader(['']+track_names,firstRow=True)
# rowSize = len(track_names)
# index=0
# while index<len(distTable):
# core.tableLine([track_names[index % rowSize]]+[str(v) for v in distTable[index:index+rowSize]])
# #for index, row in enumerate(distTable):
# # core.tableLine([track_names[index]]+[str(v) for v in row])
# core.tableFooter()
# print>>distObj, str(core)
#print>>jobFile, distMatrix.getLink('View the distance matrix for this analysis <br>')
if True: #f_matrix.shape[1] <= 100:
r_f_matrixFile = GalaxyRunSpecificFile(['f-matrix.robj'],
galaxyFn)
#', '.join([str(v) for v in row])
r.assign('f_matrix_fn', r_f_matrixFile.getDiskPath(True))
r('dput(f_matrix, f_matrix_fn)')
#r_f_matrixFile.writeTextToFile(', '.join(cls.getFlattenedMatrix(f_matrix)) + '\n\nTrack names: '+', '.join(prettyTrackNames)+'\n\nNumber of tracks: '+str(len(prettyTrackNames))+'\n\nbins: +)
#r_f_matrixFile.writeTextToFile()
#r_f_matrixFile.writeTextToFile(str(f_matrix)+'\n\n'+str(r.d))
print >> jobFile, r_f_matrixFile.getLink(
'Access the R-representation of the Feature_matrix (text-file)'
), '<br/>'
cls._clusterAndPlotDendrogram(figurepath, extra_option, 'd',
'f_matrix', prettyTrackNames)
print >> jobFile, figure.getLink(
'View the clustering tree (dendrogram) for this analysis<br>')
if True: #f_matrix.shape[1] <= 100:
#heatmap = GalaxyRunSpecificFile(['heatmap_figure.pdf'], galaxyFn)
#baseDir = os.path.dirname(heatmap.getDiskPath(True))
resDict = Results([], [], '')
resDict.setGlobalResult({
'result': {
'Matrix': f_matrix,
'Rows': np.array(track_names),
'Cols': np.array(binNames),
'Significance': None,
'RowClust': r('hr'),
'ColClust': None
}
})
header = 'View the resulting heatmap plot <br>'
baseDir = GalaxyRunSpecificFile([], galaxyFn).getDiskPath()
heatPresenter = HeatmapFromNumpyPresenter(
resDict, baseDir, header, printDimensions=False)
print >> jobFile, heatPresenter.getReference('result')
#heatmap = GalaxyRunSpecificFile(['heatmap_figure.pdf'], galaxyFn)
#heatmap_path = heatmap.getDiskPath(True)
#r.pdf(heatmap_path)
##cm.colors(256)
#r.library("gplots")
#r('heatmap(f_matrix, col=redgreen(75), distfun=function(c) dist(c, method=distanceType), hclustfun=function(c) hclust(c, method=extra_option, members=NULL),Colv=NA, scale="none", xlab="", ylab="", cexRow=0.5, cexCol=0.5, margin=c(8,10))')#Features cluster tracks
#r('dev.off()')
##print>>jobFile, r('dimnames(f_matrix)')
#print>>jobFile, heatmap.getLink('View the resulting heatmap plot <br>')
else:
print >> jobFile, 'Heatmap not generated due to large size ', f_matrix.shape
elif clusterMethod == 'K-means clustering' and extra_option != "--select--" and kmeans_alg != "--select--":
textFile = GalaxyRunSpecificFile(
['result_of_kmeans_clustering.txt'], galaxyFn)
textFilePath = textFile.getDiskPath(True)
extra_option = int(extra_option)
r.assign('kmeans_alg', kmeans_alg)
r.assign('extra_option', extra_option)
r(
'hr <- kmeans(f_matrix,extra_option,algorithm=kmeans_alg)'
) #the number of cluster is gotten from clusterMethod+ tag, instead of 3 used here
r('hr$height <- hr$height/max(hr$height)*10')
kmeans_output = open(textFilePath, 'w')
clusterSizes = r('hr$size') #size of every cluster
withinSS = r('hr$withinss')
clusters = r('hr$cluster')
for index1 in range(
extra_option
): #extra_option actually the number of clusters
#trackInCluster = [k for k,val in clusters.items() if val == index1]
trackInCluster = [
k + 1 for k, val in enumerate(clusters)
if val == index1 + 1
] #IS THIS CORRECT, I.E. SAME AS ABOVE??
print >> kmeans_output, 'Cluster %i(%s objects) : ' % (
index1 + 1, str(clusterSizes[index1]))
for name in trackInCluster:
print >> kmeans_output, name, '(This result may be a bit shaky afters some changes in rpy access)'
print >> kmeans_output, 'Sum of square error for this cluster is : ' + str(
withinSS[index1]) + '\n'
kmeans_output.close()
print >> jobFile, textFile.getLink(
'Detailed result of kmeans clustering <br>')
#cls.print_data(f_matrix, jobFile)
'''
def executeReferenceTrack(cls,
genome,
tracks,
track_names,
clusterMethod,
extra_option,
distanceType,
kmeans_alg,
galaxyFn,
regSpec,
binSpec,
numreferencetracks=None,
refTracks=None,
refFeatures=None,
yesNo=None,
howMany=None,
upFlank=None,
downFlank=None):
from proto.RSetup import r
silenceRWarnings()
jobFile = open(galaxyFn, 'w')
print >> jobFile, '<h3>Results for the "similarity of relations to other sets of genomic features" way of clustering<h3/><br/><br/>'
# print>>jobFile, 'PARAMS: ', dict(zip('genome, tracks, track_names, clusterMethod, extra_option, distanceType, kmeans_alg, regSpec, binSpec'.split(','), [repr(v)+'<br>'for v in [genome, tracks, track_names, clusterMethod, extra_option, distanceType, kmeans_alg, regSpec, binSpec]])), '<br><br>'
batchRun = GalaxyRunSpecificFile(['batch_run_job.txt'], galaxyFn)
with open(batchRun.getDiskPath(ensurePath=True), 'w') as batchFile:
print >> batchFile, '$clusterByReference', (genome, '$'.join([
':'.join(t) for t in tracks
]), ':'.join(track_names), clusterMethod, extra_option,
distanceType,
kmeans_alg, regSpec,
binSpec,
numreferencetracks,
refTracks, refFeatures,
yesNo, howMany,
upFlank, downFlank)
print >> jobFile, batchRun.getLink(
'View batch script line for this analysis<br/>')
#print>>jobFile, 'Batch script syntax for this analysis:<br>', '$clusterByReference', (genome, '$'.join([':'.join(t) for t in tracks]), ':'.join(track_names) , clusterMethod, extra_option, distanceType, kmeans_alg, regSpec, binSpec,numreferencetracks, refTracks, refFeatures, yesNo, howMany, upFlank, downFlank), '<br><br>'
#print>>jobFile, 'signature of method clusterByReference:<br>', 'clusterByReference(genome, tracksStr, track_namesStr, clusterMethod, extra_option, distanceType, kmeans_alg, regSpec, binSpec, numreferencetracks=None, refTracks=None, refFeatures=None, yesNo=None, howMany=None, upFlank=None, downFlank=None)<br><br><br>'
prettyTrackNames = [
v[-1].replace("RoadMap_", "").replace('.H3K4me1', '')
for v in tracks
]
#prettyTrackNames = [prettyPrintTrackName(v) for v in tracks]
#paramNames = ['numreferencetracks', 'refTracks', 'refFeatures', 'yesNo', 'howMany', 'upFlank', 'downFlank']
#for index, value in enumerate([numreferencetracks, refTracks, refFeatures, yesNo, howMany, upFlank, downFlank]):
# if value != None:
# print paramNames[index]+'='+ str(value),
#print ''
reftrack_names = [
] #for use in creating the heatmap (as the column names)
options = [
] #for the case using refTracks, options contains feature for every refTrack, chosen by user.
if numreferencetracks:
for i in range(int(numreferencetracks)):
ref_i = refTracks[i].split(
":"
) #name of refTrack is being used to construct the name of expanded refTrack
#refTracks.append(ref_i) #put the refTrack into refTracks list
reftrack_names.append(ref_i[-1])
temp_opt1 = 'ref' + str(i) + 'feature'
options += [] if refFeatures[i] is None else [refFeatures[i]]
if yesNo and yesNo[
i] == "Yes" and howMany and howMany[i] != '--select--':
for expan in range(int(howMany[i])):
reftrack_names.append(ref_i[-1] + '_' +
upFlank[i][expan])
upFlank = int(upFlank[i][expan])
downFlank = int(downFlank[i][expan])
withinRunId = str(i + 1) + ' expansion ' + str(expan +
1)
outTrackName = GalaxyInterface.expandBedSegmentsFromTrackNameUsingGalaxyFn(
ref_i, genome, upFlank, downFlank, galaxyFn,
withinRunId) #outTrackName is unique for run
refTracks.append(
outTrackName
) #put the expanded track into refTracks list
options.append(
options[-1]
) # use chosen feature for refTack as valid feature for the expanded
for index, track in enumerate(refTracks):
#print track, '<br>'
if isinstance(track, basestring):
track = track.split(":")
refTracks[index] = track[:-1] if track[
-1] == "-- All subtypes --" else track
if len(refTracks) > 0:
trackFormats = [
TrackInfo(genome, track).trackFormatName for track in tracks
]
trackLen = len(tracks)
refLen = len(refTracks)
f_matrix = np.zeros((trackLen, refLen))
for i in range(trackLen):
for j in range(refLen):
#print 'len(options), refLen, len(tracks), trackLen, len(trackFormats):', len(options), refLen, len(tracks), trackLen, len(trackFormats)
f_matrix[i,
j] = cls.extract_feature(genome, tracks[i],
refTracks[j], options[j],
regSpec, binSpec,
trackFormats[i])
r.assign('track_names', prettyTrackNames
) #use as track names, will be shown in clustering figure
r.assign('reftrack_names', reftrack_names)
r.assign('f_matrix', f_matrix)
r.assign('distanceType', distanceType)
r('row.names(f_matrix) <- track_names')
r('colnames(f_matrix) <- reftrack_names')
if clusterMethod == 'Hierarchical clustering' and extra_option != "--select--":
figure = GalaxyRunSpecificFile(
['cluster_tracks_result_figure.pdf'], galaxyFn)
figurepath = figure.getDiskPath(True)
#r.pdf(figurepath, 8,8)
r('d <- dist(f_matrix, method=distanceType)')
distTable = r('d')
distMatrix = GalaxyRunSpecificFile(
['distance_matrix_result.txt'], galaxyFn)
distMatrixPath = distMatrix.getDiskPath(True)
open(distMatrixPath, 'w').write(str(distTable))
print >> jobFile, distMatrix.getLink(
'View the distance matrix for this analysis <br>')
#with open(distMatrixPath,'w') as distObj:
# #distTable = d_matrix.tolist()
# core = HtmlCore()
# core.tableHeader(['']+track_names,firstRow=True)
# rowSize = len(track_names)
# index=0
# while index<len(distTable):
# core.tableLine([track_names[index % rowSize]]+[str(v) for v in distTable[index:index+rowSize]])
# core.tableFooter()
# print>>distObj, str(core)
#print>>jobFile, distMatrix.getLink('View the distance matrix for this analysis <br>')
#print r.f_matrix
#print r.d
r_f_matrixFile = GalaxyRunSpecificFile(['f-matrix.robj'],
galaxyFn)
r.assign('f_matrix_fn', r_f_matrixFile.getDiskPath(True))
r('dput(f_matrix, f_matrix_fn)')
print >> jobFile, r_f_matrixFile.getLink(
'Access the R-representation of the Feature_matrix (text-file) <br>'
),
#r_f_matrixFile = GalaxyRunSpecificFile(['f-matrix.txt'], galaxyFn)
#r_f_matrixFile.writeTextToFile(str(f_matrix)+'\n\n'+str(r.d))
#print>>jobFile, r_f_matrixFile.getLink('r.f_matrix & r.d <br>')
cls._clusterAndPlotDendrogram(figurepath, extra_option, 'd',
'f_matrix', prettyTrackNames)
#r.assign('extra_option',extra_option)
#r('hr <- hclust(d, method=extra_option, members=NULL)')
#r('hr$height <- hr$height/max(hr$height)*10')
#r('plot(hr, ylab="Distance", hang=-1)')
#
#r('dev.off()')
print >> jobFile, figure.getLink(
'View the clustering tree (dendrogram) for this analysis<br>'
)
elif clusterMethod == 'K-means clustering' and extra_option != "--select--" and kmeans_alg != "--select--":
textFile = GalaxyRunSpecificFile(
['result_of_kmeans_clustering.txt'], galaxyFn)
textFilePath = textFile.getDiskPath(True)
extra_option = int(extra_option)
r.assign('extra_option', extra_option)
r.assign('kmeans_alg', kmeans_alg)
r(
'hr <- kmeans(f_matrix,extra_option,algorithm=kmeans_alg)'
) #the number of cluster is gotten from clusterMethod+ tag, instead of 3 used here
r('hr$height <- hr$height/max(hr$height)*10')
kmeans_output = open(textFilePath, 'w')
clusterSizes = r('hr$size') #size of every cluster
withinSS = r('hr$withinss')
clusters = np.array(
r('hr$cluster')
) #convert to array in order to handle the index more easily
track_names = np.array(track_names)
for index1 in range(
extra_option
): #extra_option actually the number of clusters
trackInCluster = [
k for k, val in clusters.items() if val == index1
]
print >> kmeans_output, 'Cluster %i(%s objects) : ' % (
index1 + 1, str(clusterSizes[index1]))
for name in trackInCluster:
print >> kmeans_output, name
print >> kmeans_output, 'Sum of square error for this cluster is : ' + str(
withinSS[index1]) + '\n'
kmeans_output.close()
print >> jobFile, textFile.getLink(
'Detailed result of kmeans clustering <br>')
#heatmap = GalaxyRunSpecificFile(['heatmap_figure.pdf'], galaxyFn)
#baseDir = os.path.dirname(heatmap.getDiskPath(True))
##r.png(heatmap_path, width=800, height=700)
resDict = Results([], [], 'ClusTrack')
resDict.setGlobalResult({
'result': {
'Matrix': f_matrix,
'Rows': np.array(track_names),
'Cols': np.array(reftrack_names),
'Significance': None,
'RowClust': r('hr'),
'ColClust': None
}
})
header = 'Heatmap of Feature matrix for "similarity of positional distribution along the genome" '
baseDir = GalaxyRunSpecificFile([], galaxyFn).getDiskPath()
heatPresenter = HeatmapFromNumpyPresenter(resDict,
baseDir,
header,
printDimensions=False)
print >> jobFile, heatPresenter.getReference('result')
#r.pdf(heatmap_path)
#r.library("gplots")
#r('heatmap(f_matrix, col=redgreen(75), Colv=NA, scale="none", xlab="", ylab="", margins=c(10,10))')#Features cluster tracks
#r('dev.off()')
#print>>jobFile, heatmap.getLink('View the resulting heatmap plot <br>')
#cls.print_data(f_matrix, jobFile)
else:
print 'Have to specify a set of refTracks'
def _emptyResults(self):
return Results(["Track 1"], ["Track 2"], self._statClass.__name__)
def _emptyResults(self):
return Results(self._track.trackName, self._track2.trackName \
if self._track2 is not None else [], self._statClass.__name__)