def mergeAlignments(daddyAlignmentFile, childAlignmentFile, childSeq, childSeq2, daddySeqNo, childSeqNo, newAlignmentFile, nodeLabels):
childGapColumn = ['-']*childSeqNo
topDaddyGapColumn = ['-']*childSeq
bottomDaddyGapColumn = ['-']*(daddySeqNo-(childSeq+1))
totalSeqNo = daddySeqNo -1 + childSeqNo
outputFiles, outputIters = getOpenSeqFiles(totalSeqNo, getTempFile)
childIter = multiFastaRead(childAlignmentFile)
def nextChild():
try:
return childIter.next()
except StopIteration:
return None
childColumn = nextChild()
for daddyColumn in multiFastaRead(daddyAlignmentFile):
if daddyColumn[childSeq] != '-':
while childColumn[childSeq2] == '-':
outputMergedColumn(outputIters, topDaddyGapColumn, childColumn, bottomDaddyGapColumn)
childColumn = nextChild()
outputMergedColumn(outputIters, daddyColumn[:childSeq], childColumn, daddyColumn[childSeq+1:])
childColumn = nextChild()
else:
outputMergedColumn(outputIters, daddyColumn[:childSeq], childGapColumn, daddyColumn[childSeq+1:])
while childColumn != None:
outputMergedColumn(outputIters, topDaddyGapColumn, childColumn, bottomDaddyGapColumn)
childColumn = nextChild()
closeSeqIterators(outputIters, totalSeqNo)
concatanateSeqFiles(outputFiles, newAlignmentFile, totalSeqNo, nodeLabels[0:totalSeqNo])
removeSeqFiles(outputFiles, totalSeqNo)
def countGaplessColumns(alignment):
gapless = 0
total = 0
for column in multiFastaRead(alignment):
if '-' not in column:
gapless += 1
total += 1
return gapless, total
def countGaplessColumns(alignment):
gapless = 0
total = 0
for column in multiFastaRead(alignment):
if '-' not in column:
gapless += 1
total += 1
return gapless, total
def getGaplessAlignment(alignment, seqNo):
outputAlignment = getTempFile()
outputFiles, outputIters = getOpenSeqFiles(seqNo, getTempFile)
for column in multiFastaRead(alignment):
if '-' not in column:
for i in xrange(0, seqNo):
outputIters[i].write(column[i])
closeSeqIterators(outputIters, seqNo)
concatanateSeqFiles(outputFiles, outputAlignment, seqNo, [ str(i) for i in xrange(0, seqNo) ])
removeSeqFiles(outputFiles, seqNo)
return outputAlignment
def extractSubAlignment(alignmentFile, startSeq, endSeq, newAlignmentFile):
seqNo = endSeq-startSeq
outputFiles, outputIters = getOpenSeqFiles(seqNo, getTempFile)
for column in multiFastaRead(alignmentFile):
for i in xrange(startSeq, endSeq):
if column[i] != '-':
for j in xrange(startSeq, endSeq):
outputIters[j-startSeq].write(column[j])
break
closeSeqIterators(outputIters, seqNo)
concatanateSeqFiles(outputFiles, newAlignmentFile, seqNo, [ str(i) for i in xrange(startSeq, endSeq) ])
removeSeqFiles(outputFiles, seqNo)
def getGaplessAlignment(alignment, seqNo):
outputAlignment = getTempFile()
outputFiles, outputIters = getOpenSeqFiles(seqNo, getTempFile)
for column in multiFastaRead(alignment):
if '-' not in column:
for i in xrange(0, seqNo):
outputIters[i].write(column[i])
closeSeqIterators(outputIters, seqNo)
concatanateSeqFiles(outputFiles, outputAlignment, seqNo,
[str(i) for i in xrange(0, seqNo)])
removeSeqFiles(outputFiles, seqNo)
return outputAlignment
def mergeAlignments(daddyAlignmentFile, childAlignmentFile, childSeq,
childSeq2, daddySeqNo, childSeqNo, newAlignmentFile,
nodeLabels):
childGapColumn = ['-'] * childSeqNo
topDaddyGapColumn = ['-'] * childSeq
bottomDaddyGapColumn = ['-'] * (daddySeqNo - (childSeq + 1))
totalSeqNo = daddySeqNo - 1 + childSeqNo
outputFiles, outputIters = getOpenSeqFiles(totalSeqNo, getTempFile)
childIter = multiFastaRead(childAlignmentFile)
def nextChild():
try:
return childIter.next()
except StopIteration:
return None
childColumn = nextChild()
for daddyColumn in multiFastaRead(daddyAlignmentFile):
if daddyColumn[childSeq] != '-':
while childColumn[childSeq2] == '-':
outputMergedColumn(outputIters, topDaddyGapColumn, childColumn,
bottomDaddyGapColumn)
childColumn = nextChild()
outputMergedColumn(outputIters, daddyColumn[:childSeq],
childColumn, daddyColumn[childSeq + 1:])
childColumn = nextChild()
else:
outputMergedColumn(outputIters, daddyColumn[:childSeq],
childGapColumn, daddyColumn[childSeq + 1:])
while childColumn != None:
outputMergedColumn(outputIters, topDaddyGapColumn, childColumn,
bottomDaddyGapColumn)
childColumn = nextChild()
closeSeqIterators(outputIters, totalSeqNo)
concatanateSeqFiles(outputFiles, newAlignmentFile, totalSeqNo,
nodeLabels[0:totalSeqNo])
removeSeqFiles(outputFiles, totalSeqNo)
def extractSubAlignment(alignmentFile, startSeq, endSeq, newAlignmentFile):
seqNo = endSeq - startSeq
outputFiles, outputIters = getOpenSeqFiles(seqNo, getTempFile)
for column in multiFastaRead(alignmentFile):
for i in xrange(startSeq, endSeq):
if column[i] != '-':
for j in xrange(startSeq, endSeq):
outputIters[j - startSeq].write(column[j])
break
closeSeqIterators(outputIters, seqNo)
concatanateSeqFiles(outputFiles, newAlignmentFile, seqNo,
[str(i) for i in xrange(startSeq, endSeq)])
removeSeqFiles(outputFiles, seqNo)
def stitchReconstruct(seqNo, inputSeqFiles, treeString, outputFile, outputScoreFile, inputAlignmentFile, alignerArgs):
startTime = time.time() #epoch time in seconds
logger.info("Starting Stitcher")
reconstructionPrefix = alignerArgs.RECONSTRUCTION_PREFIX
if alignerArgs.FAST_SETTING:
reconstructionArgs = alignerArgs.RECONSTRUCTION_ARGS_FAST
else:
reconstructionArgs = alignerArgs.RECONSTRUCTION_ARGS
cautiousArgs = alignerArgs.CAUTIOUS_ARGS
alignmentChunkMaxSeqSize = alignerArgs.ALIGNMENT_CHUNK_MAX_COLUMN_SIZE
viterbiAlignmentColumnGap = alignerArgs.VITERBI_ALIGNMENT_COLUMN_GAP
#parse tree
binaryTree = newickTreeParser(treeString)
binaryTree_depthFirstNumbers(binaryTree)
logger.info("Newick tree read : %s " % printBinaryTree(binaryTree, True))
labels = binaryTree_nodeNames(binaryTree)
leafLabels = [ labels[i] for i in xrange(0, len(labels)) if (i%2) == 0]
#load alignment iterator
alignmentReader = multiFastaRead(inputAlignmentFile, lambda x : x)
#number of sequences, including ancestors
nodeNumber = binaryTree.traversalID.midEnd
assert nodeNumber == seqNo * 2 - 1
#create output files
outputFiles, outputIterators = getOpenSeqFiles(nodeNumber, getTempFile)
#while has chunk
previousAlignment = []
alignmentSeqs, alignmentFile, end = getNextAlignmentChunk(previousAlignment, alignmentReader, alignmentChunkMaxSeqSize, seqNo, leafLabels)
tempTreeStatesFile = getTempFile()
loopOptions = " "
logger.info("Starting main loop")
characterFrequenciesString = " ".join([ str(i) for i in alignerArgs.EXPECTED_CHARACTER_FREQUENCIES ])
while alignmentSeqs != None:
if(end):
viterbiAlignmentColumnGap = 0
tempAncestorFile = getTempFile()
tempScoreFile = getTempFile()
command = "%s -b '%s' -c %s -a %s -u %s -s %s %s %s -d %s -n %s -x %s " % (reconstructionPrefix, treeString, alignmentFile, \
" ".join(alignmentSeqs), tempTreeStatesFile, \
viterbiAlignmentColumnGap, loopOptions, reconstructionArgs, tempAncestorFile, characterFrequenciesString, tempScoreFile)
logger.info("Calling Ortheus with : %s", command)
exitValue = os.system(command)
if exitValue != 0:
logger.info("Something went wrong calling Ortheus : %i ", exitValue)
#if exitValue != 73:
# logger.info("Unrecognised issue, so am exiting to be cautious")
# sys.exit(1)
logger.info("Going to retry with caution settings")
command = "%s -b '%s' -c %s -a %s -u %s -s %s %s %s -d %s -x %s" % (reconstructionPrefix, treeString, alignmentFile, \
" ".join(alignmentSeqs), tempTreeStatesFile, \
viterbiAlignmentColumnGap, loopOptions, cautiousArgs, tempAncestorFile, tempScoreFile)
logger.info("Calling Ortheus with : %s", command)
if os.system(command):
logger.info("Already tried caution, so have to go")
sys.exit(1)
logger.info("Completed reconstruction of chunk")
appendScore(tempScoreFile, outputScoreFile)
os.remove(tempScoreFile)
loopOptions = " -t " + tempTreeStatesFile
tempAncestorFastaOffsets = getMultiFastaOffsets(tempAncestorFile)
previousAlignment = removeFromLeft(multiFastaRead(tempAncestorFile, lambda x : x, tempAncestorFastaOffsets), previousAlignment, nodeNumber, seqNo)
appendToAlignment(multiFastaRead(tempAncestorFile, lambda x : x, tempAncestorFastaOffsets), outputIterators, nodeNumber)
logger.info("Added reconstructed chunk to complete alignment")
os.remove(tempAncestorFile)
removeSeqFiles(alignmentSeqs, seqNo)
os.remove(alignmentFile)
logger.info("Cleaned up at end of loop")
alignmentSeqs, alignmentFile, end = getNextAlignmentChunk(previousAlignment, alignmentReader, alignmentChunkMaxSeqSize, seqNo, leafLabels)
logger.info("Finished main loop")
#load into single output file
closeSeqIterators(outputIterators, nodeNumber)
concatanateSeqFiles(outputFiles, outputFile, nodeNumber, labels)
logger.info("Written out alignment to single file")
#clean up
os.remove(tempTreeStatesFile)
removeSeqFiles(outputFiles, nodeNumber)
logger.info("Cleaned up final files")
logger.info("Finished, total time taken for stitcher: %s (seconds)" % (time.time()-startTime))