def createMultiGroupProfile(self, groupNames, parentHeading,
profileHeading, metadata,
unclassifiedTreatment):
multiGroupProfile = MultiGroupProfile()
multiGroupProfile.groupNames = sorted(groupNames)
# get depth of hierarchical levels of interest
if parentHeading == 'Entire sample':
parentDepth = 0
else:
parentDepth = self.hierarchyHeadings.index(parentHeading) + 1
profileDepth = self.hierarchyHeadings.index(profileHeading) + 1
multiGroupProfile.hierarchyHeadings = self.hierarchyHeadings[
0:profileDepth]
# get list of samples in each group for samples of interest
multiGroupProfile.samplesInGroups = []
samples = []
multiGroupProfile.smallestGroup = sys.maxint
for groupName in multiGroupProfile.groupNames:
samplesInGroup = list(
set(self.groupDict[groupName]).intersection(
metadata.activeSamples))
sortedSampleNames = sorted(samplesInGroup)
multiGroupProfile.samplesInGroups.append(sortedSampleNames)
samples += sortedSampleNames
if len(sortedSampleNames) < multiGroupProfile.smallestGroup:
multiGroupProfile.smallestGroup = len(sortedSampleNames)
# get counts for all samples
leafNodes = self.getLeafNodes()
# traverse up tree from each leaf node
parentSeqDict = {}
for leaf in leafNodes:
curDepth = len(self.hierarchyHeadings)
curNode = leaf
hierarchy = []
bRemoveUnclassified = False
while curNode != None:
if not curNode.isRoot() and curDepth <= profileDepth:
hierarchy.append(curNode.name)
# add profile level information
if curDepth == profileDepth:
if 'unclassified' in curNode.name.lower():
if unclassifiedTreatment == 'Remove unclassified reads':
bRemoveUnclassified = True
break
elif unclassifiedTreatment == 'Use only for calculating frequency profiles':
bRemoveUnclassified = True
if bRemoveUnclassified == False:
name = curNode.name
# remove ' - #' if feature is being calculated relative to the entire sample
bTruncatedName = False
if curNode.isLeaf() and parentDepth == 0:
if name.rfind(' - #') != -1:
name = name[0:name.rfind(' - #')]
bTruncatedName = True
profileEntry = multiGroupProfile.profileDict.get(name)
if bTruncatedName == True and profileEntry != None:
bRemoveUnclassified = True
break
if profileEntry == None:
profileEntry = GroupProfileEntry()
profileEntry.featureCounts = [0] * len(samples)
multiGroupProfile.profileDict[
curNode.name] = profileEntry
col = 0
for sampleName in samples:
profileEntry.featureCounts[col] += leaf.countData[
sampleName]
col += 1
# add parent level information
if curDepth == parentDepth:
sequences = parentSeqDict.get(curNode.name)
if sequences == None:
sequences = [0] * len(samples)
parentSeqDict[curNode.name] = sequences
col = 0
for sampleName in samples:
sequences[col] += leaf.countData[sampleName]
col += 1
if bRemoveUnclassified == False:
profileEntry.parentCounts = sequences
curDepth -= 1
curNode = curNode.parent
if bRemoveUnclassified == False:
hierarchy.reverse()
profileEntry.hierarchy = hierarchy
multiGroupProfile.numParentCategories = len(parentSeqDict)
multiGroupProfile.setActiveGroups(self.groupActive)
return multiGroupProfile
def createGroupProfile(self, groupName1, groupName2, parentHeading,
profileHeading, metadata, unclassifiedTreatment):
groupProfile = GroupProfile()
if groupName1 == '' or groupName2 == '':
return groupProfile
groupProfile.groupName1 = groupName1
groupProfile.groupName2 = groupName2
# get depth of hierarchical levels of interest
if parentHeading == 'Entire sample':
parentDepth = 0
else:
parentDepth = self.hierarchyHeadings.index(parentHeading) + 1
profileDepth = self.hierarchyHeadings.index(profileHeading) + 1
groupProfile.hierarchyHeadings = self.hierarchyHeadings[0:profileDepth]
# get list of samples in each group for samples of interest
samplesInGroup1 = list(
set(self.groupDict[groupName1]).intersection(
metadata.activeSamples))
if groupName2 != '<All other samples>':
samplesInGroup2 = list(
set(self.groupDict[groupName2]).intersection(
metadata.activeSamples))
else:
samplesInGroup2 = set([])
for groupName in self.groupDict:
if groupName != groupName1:
samplesInGroup2 = samplesInGroup2.union(
set(self.groupDict[groupName]).intersection(
metadata.activeSamples))
samplesInGroup2 = list(samplesInGroup2)
groupProfile.samplesInGroup1 = sorted(samplesInGroup1)
groupProfile.samplesInGroup2 = sorted(samplesInGroup2)
samples = groupProfile.samplesInGroup1 + groupProfile.samplesInGroup2
# get counts for all samples
leafNodes = self.getLeafNodes()
# traverse up tree from each leaf node
parentSeqDict = {}
for leaf in leafNodes:
curDepth = len(self.hierarchyHeadings)
curNode = leaf
hierarchy = []
bRemoveUnclassified = False
while curNode != None:
if not curNode.isRoot() and curDepth <= profileDepth:
hierarchy.append(curNode.name)
# add profile level information
if curDepth == profileDepth:
if 'unclassified' in curNode.name.lower():
if unclassifiedTreatment == 'Remove unclassified reads':
bRemoveUnclassified = True
break
elif unclassifiedTreatment == 'Use only for calculating frequency profiles':
bRemoveUnclassified = True
if bRemoveUnclassified == False:
name = curNode.name
# remove ' - #' if feature is being calculated relative to the entire sample
bTruncatedName = False
if curNode.isLeaf() and parentDepth == 0:
if name.rfind(' - #') != -1:
name = name[0:name.rfind(' - #')]
bTruncatedName = True
profileEntry = groupProfile.profileDict.get(name)
if bTruncatedName == True and profileEntry != None:
bRemoveUnclassified = True
break
if profileEntry == None:
profileEntry = GroupProfileEntry()
profileEntry.featureCounts = [0] * len(samples)
groupProfile.profileDict[name] = profileEntry
col = 0
for sampleName in samples:
profileEntry.featureCounts[col] += leaf.countData[
sampleName]
col += 1
# add parent level information
if curDepth == parentDepth:
sequences = parentSeqDict.get(curNode.name)
if sequences == None:
sequences = [0] * len(samples)
parentSeqDict[curNode.name] = sequences
col = 0
for sampleName in samples:
sequences[col] += leaf.countData[sampleName]
col += 1
if bRemoveUnclassified == False:
profileEntry.parentCounts = sequences
curDepth -= 1
curNode = curNode.parent
if bRemoveUnclassified == False:
hierarchy.reverse()
profileEntry.hierarchy = hierarchy
groupProfile.numParentCategories = len(parentSeqDict)
return groupProfile
