def callSAIWorkers(fastqDir, fastaPath, numThreads = 12):
#contained all fastas against which to map the fastqs
fastaFilePaths = workerForBam.getFastaFilePaths(fastaPath)
print fastaFilePaths
#contained all fastqs
fastqFilePaths = workerForBam.getFastqFilePaths(fastqDir)
print fastqFilePaths
#get number of processors
numProcs = multiprocessing.cpu_count()
print 'number of cpus: ', numProcs
#compute number of jobs
numJobs = len(fastqFilePaths)*len(fastaFilePaths)
#initialize pool with number of possible jobs
pool = Pool(processes=min(numJobs, numProcs))
workersArgs = []
#for each read and fasta create a job
for fastaFilePath in fastaFilePaths:
for fastqPath in fastqFilePaths:
workersArgs.append((fastaFilePath, fastqPath, numThreads))
results = pool.map(callSAIWorker, workersArgs)
pool.close()
pool.join()
return results
def writeCombineBAMJobsFromSAI(outDir, fastqDir, fastaPath, lockDirPath):
combinedBAMJobsName = "combinedBAMFrmSingleSAIJob.jobs"
combinedBAMJobsPath = os.path.join(outDir, combinedBAMJobsName)
tools = workerForBam.getToolsDict()
# contained all fastas against which to map the fastqs
fastaFilePaths = workerForBam.getFastaFilePaths(fastaPath)
# contained all fastqs
fastqFilePaths = workerForBam.getFastqFilePaths(fastqDir)
print "fastaFilePaths: ", fastaFilePaths
with open(combinedBAMJobsPath, "w") as combinedBAMJobsFile:
for fastqPath in fastqFilePaths:
for fastaFilePath in fastaFilePaths:
workerForBam.writeSAIIToBAMJob(combinedBAMJobsFile, fastaFilePath, fastqPath, lockDirPath, tools)
return combinedBAMJobsPath
def writeCombineSAIJobs(outDir, fastqDir, fastaPath, lockDirPath, numThreads):
combinedSAIJobsName = 'combinedSAIJob.jobs'
combinedSAIJobsPath = os.path.join(outDir, combinedSAIJobsName)
tools = workerForBam.getToolsDict()
#contained all fastas against which to map the fastqs
fastaFilePaths = workerForBam.getFastaFilePaths(fastaPath)
#contained all fastqs
fastqFilePaths = workerForBam.getFastqFilePaths(fastqDir)
print 'fastaFilePaths: ', fastaFilePaths
with open(combinedSAIJobsPath, 'w') as combinedSAIJobsFile:
for fastqPath in fastqFilePaths:
for fastaFilePath in fastaFilePaths:
workerForBam.writeSAIJob(combinedSAIJobsFile, fastaFilePath,\
fastqPath, lockDirPath, tools,\
numThreads)
return combinedSAIJobsPath
def callSAIWorkers(fastqDir, fastaPath, numThreads = 12):
#contained all fastas against which to map the fastqs
fastaFilePaths = workerForBam.getFastaFilePaths(fastaPath)
print fastaFilePaths
#contained all fastqs
fastqFilePaths = workerForBam.getFastqFilePaths(fastqDir)
print fastqFilePaths
#compute number of jobs
numJobs = len(fastqFilePaths)*len(fastaFilePaths)
print 'number of jobs: ', numJobs
sys.stdout.flush()
#for each read and fasta create a job
for fastaFilePath in fastaFilePaths:
for fastqPath in fastqFilePaths:
result = callSAIWorker((fastaFilePath, fastqPath, numThreads))
print result, (fastaFilePath, fastqPath, numThreads)
sys.stdout.flush()