def handlePostPileupMerge(self, reads):
'''
This function compares a group of alignments and decides which one to keep
@param reads - a set of reads with the same name to be compared using pileup
'''
avgSum = 0
[pairs, singles] = MergeImprove.pairReads(reads,
MergeImprove.PILEUP_HI_TAG)
if len(pairs) != 0:
bestAvgPileup = -1
bestPairs = []
for pair in pairs:
[tot1, bases1] = self.calcPileupStats(pair[0])
[tot2, bases2] = self.calcPileupStats(pair[1])
avgPileup = float(tot1 + tot2) / (bases1 + bases2)
avgSum += avgPileup
if avgPileup > bestAvgPileup:
bestAvgPileup = avgPileup
bestPairs = []
if avgPileup == bestAvgPileup:
bestPairs.append(pair)
#stats
if len(bestPairs) == 1:
MergeImprove.setTag(bestPairs[0][0],
MergeImprove.CHOICE_TYPE_TAG, 'P')
MergeImprove.setTag(bestPairs[0][1],
MergeImprove.CHOICE_TYPE_TAG, 'P')
#save pileup stats
if (bestAvgPileup == 0):
self.percentageChoice[0] += 2
else:
self.percentageChoice[int(100 * bestAvgPileup /
avgSum)] += 2
#save one of the best pileup pairs
self.saveRandomPair(bestPairs)
else:
#do this over singles
bestAvgPileup = {}
bestReads = {}
avgSum = {False: 0, True: 0}
for read in singles:
#if there's nothing yet for this sequence, set it's best as -1 so it gets overwritten below
isFirst = MergeImprove.isFlagSet(
read.flag, MergeImprove.FIRST_SEGMENT_FLAG)
if not bestAvgPileup.has_key(isFirst):
bestAvgPileup[isFirst] = -1
bestReads[isFirst] = []
#get the pileup calculation
[tot, bases] = self.calcPileupStats(read)
avgPileup = float(tot) / bases
avgSum[isFirst] += avgPileup
#if it's better, keep it
if avgPileup > bestAvgPileup[isFirst]:
bestAvgPileup[isFirst] = avgPileup
bestReads[isFirst] = []
if avgPileup == bestAvgPileup[isFirst]:
bestReads[isFirst].append(read)
#save the best from each end
for end in bestReads:
brs = bestReads[end]
if len(brs) == 1:
MergeImprove.setTag(brs[0], MergeImprove.CHOICE_TYPE_TAG,
'P')
if bestAvgPileup[end] == 0:
self.percentageChoice[0] += 1
else:
self.percentageChoice[int(100 * bestAvgPileup[end] /
avgSum[end])] += 1
self.saveRandomSingle(brs)
def handlePostPileupMerge(self, reads):
'''
This function compares a group of alignments and decides which one to keep
@param reads - a set of reads with the same name to be compared using pileup
'''
avgSum = 0
[pairs, singles] = MergeImprove.pairReads(reads, MergeImprove.PILEUP_HI_TAG)
if len(pairs) != 0:
bestAvgPileup = -1
bestPairs = []
for pair in pairs:
[tot1, bases1] = self.calcPileupStats(pair[0])
[tot2, bases2] = self.calcPileupStats(pair[1])
avgPileup = float(tot1+tot2)/(bases1+bases2)
avgSum += avgPileup
if avgPileup > bestAvgPileup:
bestAvgPileup = avgPileup
bestPairs = []
if avgPileup == bestAvgPileup:
bestPairs.append(pair)
#stats
if len(bestPairs) == 1:
MergeImprove.setTag(bestPairs[0][0], MergeImprove.CHOICE_TYPE_TAG, 'P')
MergeImprove.setTag(bestPairs[0][1], MergeImprove.CHOICE_TYPE_TAG, 'P')
#save pileup stats
if(bestAvgPileup == 0):
self.percentageChoice[0] += 2
else:
self.percentageChoice[int(100*bestAvgPileup/avgSum)] += 2
#save one of the best pileup pairs
self.saveRandomPair(bestPairs)
else:
#do this over singles
bestAvgPileup = {}
bestReads = {}
avgSum = {False: 0, True: 0}
for read in singles:
#if there's nothing yet for this sequence, set it's best as -1 so it gets overwritten below
isFirst = MergeImprove.isFlagSet(read.flag, MergeImprove.FIRST_SEGMENT_FLAG)
if not bestAvgPileup.has_key(isFirst):
bestAvgPileup[isFirst] = -1
bestReads[isFirst] = []
#get the pileup calculation
[tot, bases] = self.calcPileupStats(read)
avgPileup = float(tot)/bases
avgSum[isFirst] += avgPileup
#if it's better, keep it
if avgPileup > bestAvgPileup[isFirst]:
bestAvgPileup[isFirst] = avgPileup
bestReads[isFirst] = []
if avgPileup == bestAvgPileup[isFirst]:
bestReads[isFirst].append(read)
#save the best from each end
for end in bestReads:
brs = bestReads[end]
if len(brs) == 1:
MergeImprove.setTag(brs[0], MergeImprove.CHOICE_TYPE_TAG, 'P')
if bestAvgPileup[end] == 0:
self.percentageChoice[0] += 1
else:
self.percentageChoice[int(100*bestAvgPileup[end]/avgSum[end])] += 1
self.saveRandomSingle(brs)
