def testMulticoreAnnotate(self):
"""Test a (too) simple annotating exercise from GAF on 2 cores"""
gafDatasource = TestUtils.createGafDatasourceProxy(self.config)
# Test pickling
dump(gafDatasource, file('out/testGAFPickle.pkl', 'w'))
m1 = MutationDataFactory.default_create()
m1.chr = '3'
m1.start = '178866811'
m1.end = '178866811'
m1.ref_allele = "A"
m1.alt_allele = "C"
m1.build = "hg19"
m2 = MutationDataFactory.default_create()
m2.chr = '3'
m2.start = '178866812'
m2.end = '178866812'
m2.ref_allele = "A"
m2.alt_allele = "C"
m2.build = "hg19"
p = LoggingPool(processes=2)
result = p.map(annotate_mutation_global, [(gafDatasource, m1),
(gafDatasource, m2)])
p.close()
p.join()
for r in result:
self.assertTrue("transcript_id" in r.keys())
self.assertTrue("gene" in r.keys())
self.assertTrue(r["gene"] == "PIK3CA")
self.assertTrue(result[0].start != result[1].start)
def testMulticoreAnnotate(self):
"""Test a (too) simple annotating exercise from GAF on 2 cores"""
gafDatasource = TestUtils.createGafDatasourceProxy(self.config)
# Test pickling
dump(gafDatasource, file('out/testGAFPickle.pkl','w'))
m1 = MutationDataFactory.default_create()
m1.chr = '3'
m1.start = '178866811'
m1.end = '178866811'
m1.ref_allele = "A"
m1.alt_allele = "C"
m1.build = "hg19"
m2 = MutationDataFactory.default_create()
m2.chr = '3'
m2.start = '178866812'
m2.end = '178866812'
m2.ref_allele = "A"
m2.alt_allele = "C"
m2.build = "hg19"
p = LoggingPool(processes=2)
result = p.map(annotate_mutation_global, [(gafDatasource, m1), (gafDatasource, m2)])
p.close()
p.join()
for r in result:
self.assertTrue("transcript_id" in r.keys())
self.assertTrue("gene" in r.keys())
self.assertTrue(r["gene"] == "PIK3CA")
self.assertTrue(result[0].start != result[1].start)
def testMulticoreAnnotateFromChunkedFile(self):
#TODO: Add unit test that Mutation data is pickle-able
inputFile = "testdata/maflite/Patient0.snp.maf.txt"
outputFile = "out/testGAFMulticorePatient0.snp.maf.txt"
chunkSize = 200
numChunks = 4
gafDatasource = TestUtils.createGafDatasourceProxy(self.config)
ic = MafliteInputMutationCreator(inputFile)
oc = SimpleOutputRenderer(outputFile)
# createChunks
muts = ic.createMutations()
allAnnotatedChunksFlat = []
are_mutations_remaining = True
p = LoggingPool(processes=numChunks)
while are_mutations_remaining:
chunks = []
for j in xrange(0, numChunks):
chunk = []
for i in xrange(0, chunkSize):
try:
chunk.append(muts.next())
except StopIteration:
are_mutations_remaining = False
break
chunks.append((chunk, gafDatasource))
annotatedChunks = p.map(annotate_mutations_global, chunks)
annotatedChunksFlat = self._flattenChunks(annotatedChunks)
allAnnotatedChunksFlat.append(annotatedChunksFlat)
p.close()
p.join()
annotatedMuts = chain.from_iterable(allAnnotatedChunksFlat)
ctr = 0
oc.renderMutations(annotatedMuts, Metadata())
tsvReader = GenericTsvReader(outputFile)
for line in tsvReader:
ctr += 1
self.assertTrue(ctr == 730,
"Should have read 730 variants, but read " + str(ctr))
def testMulticoreAnnotateFromChunkedFile(self):
#TODO: Add unit test that Mutation data is pickle-able
inputFile = "testdata/maflite/Patient0.snp.maf.txt"
outputFile = "out/testGAFMulticorePatient0.snp.maf.txt"
chunkSize = 200
numChunks = 4
gafDatasource = TestUtils.createGafDatasourceProxy(self.config)
ic = MafliteInputMutationCreator(inputFile)
oc = SimpleOutputRenderer(outputFile)
# createChunks
muts = ic.createMutations()
allAnnotatedChunksFlat = []
are_mutations_remaining = True
p = LoggingPool(processes=numChunks)
while are_mutations_remaining:
chunks = []
for j in xrange(0, numChunks):
chunk = []
for i in xrange(0, chunkSize):
try:
chunk.append(muts.next())
except StopIteration:
are_mutations_remaining = False
break
chunks.append((chunk, gafDatasource))
annotatedChunks = p.map(annotate_mutations_global, chunks)
annotatedChunksFlat = self._flattenChunks(annotatedChunks)
allAnnotatedChunksFlat.append(annotatedChunksFlat)
p.close()
p.join()
annotatedMuts = chain.from_iterable(allAnnotatedChunksFlat)
ctr = 0
oc.renderMutations(annotatedMuts, Metadata())
tsvReader = GenericTsvReader(outputFile)
for line in tsvReader:
ctr += 1
self.assertTrue(ctr == 730, "Should have read 730 variants, but read " + str(ctr))