def testSampleNameSelectorWithMaf(self):
input = MafliteInputMutationCreator("testdata/maflite/tiny_maflite.maf.txt")
first_mut = next(input.createMutations())
s = SampleNameSelector(first_mut)
for mut in input.createMutations():
self.assertEqual("Patient0-Normal-Patient0-Tumor", s.getSampleName(mut))
self.assertEqual(s.getAnnotationSource(),"OUTPUT")
self.assertEqual(s.getOutputAnnotationName(), MutUtils.SAMPLE_NAME_ANNOTATION_NAME)
def testSampleNameSelectorWithVCF(self):
input = VcfInputMutationCreator("testdata/vcf/example.1row.vcf")
first_mut = next(input.createMutations())
s = SampleNameSelector(first_mut)
expected = ["NA 00001", "NA 00002", "NA 00003"]
for mut in input.createMutations():
self.assertIn(s.getSampleName(mut), expected)
self.assertEqual(s.getAnnotationSource(), "INPUT")
self.assertEquals(s.getOutputAnnotationName(), "sample_name")
def testSampleNameSelectorWithVCF(self):
input = VcfInputMutationCreator("testdata/vcf/example.1row.vcf")
first_mut = next(input.createMutations())
s = SampleNameSelector(first_mut)
expected = ["NA 00001", "NA 00002", "NA 00003"]
for mut in input.createMutations():
self.assertIn(s.getSampleName(mut), expected)
self.assertEqual(s.getAnnotationSource(), "INPUT")
self.assertEquals(s.getOutputAnnotationName(), "sample_name")
def testSampleNameSelectorWithMaf(self):
input = MafliteInputMutationCreator(
"testdata/maflite/tiny_maflite.maf.txt")
first_mut = next(input.createMutations())
s = SampleNameSelector(first_mut)
for mut in input.createMutations():
self.assertEqual("Patient0-Normal-Patient0-Tumor",
s.getSampleName(mut))
self.assertEqual(s.getAnnotationSource(), "OUTPUT")
self.assertEqual(s.getOutputAnnotationName(),
MutUtils.SAMPLE_NAME_ANNOTATION_NAME)
def __init__(self, mutations):
"""
Initialize an new queue with a MutationData iterator
:param mutations: any MutationData producing Iterator
"""
self.mutations = more_itertools.peekable(mutations)
self.sns = SampleNameSelector(self.mutations.peek())
self.queue = collections.defaultdict(list)
self.indel_queue = []
self.last = 0
self.logger = logging.getLogger(__name__)
self.warned_about_order = False
def __init__(self, mutations, mutation_data_factory):
"""
Initialize an new queue with a MutationData iterator
:param mutations: any MutationData producing Iterator
:param mutation_data_factory: a MutationDataFactory to be used to produce new mutations for the ONPs
"""
self.mutations = more_itertools.peekable(mutations)
self.sns = SampleNameSelector(self.mutations.peek())
self.queue = collections.defaultdict(list)
self.indel_queue = []
self.last = 0
self.logger = logging.getLogger(__name__)
self.warned_about_order = False
self._mutation_data_factory = mutation_data_factory
def _writeMuts2Tsv(self, muts, path):
"""
Given a mutation generator, this methods writes a tab separated file for all mutations in the mutation
generator. In addition, this method computes the appropriate sample name in scenarios where the mutation is
missing sample name annotation. It also computes a list of all chromosomes and sample names contained within
the generator.
:param path: temporary filename
:param muts: generator object with mutations
"""
sampleNames = set()
chroms = set()
writer = None
# create a temporary file to write tab-separated file
tempTsvFile = tempfile.NamedTemporaryFile(dir=path, delete=False)
self.logger.debug("Creating intermediate tsv file at %s" % tempTsvFile.name)
mutAttributeNames = []
sampleNameSelector = SampleNameSelector(self.mutation,
configFile=self.configTable.getConfigFilename(),
section="OTHER")
with open(tempTsvFile.name, 'w') as fptr:
ctr = 0
sampleNameAnnotationName = sampleNameSelector.getOutputAnnotationName()
sampleNameSource = sampleNameSelector.getAnnotationSource()
for mut in muts:
if len(mutAttributeNames) == 0:
mutAttributeNames = mut.getAttributeNames()
sampleName = sampleNameSelector.getSampleName(mut)
if sampleName is not None:
if mut.get(sampleNameAnnotationName, None) is None:
mut.createAnnotation(sampleNameAnnotationName, sampleName, sampleNameSource)
sampleNames.add(sampleName)
# Parse chromosome
chroms.add(mut.chr)
updated_start, updated_ref_allele, updated_alt_allele = MutUtils.retrieveMutCoordinatesForRendering(mut)
mut.ref_allele = updated_ref_allele
mut.alt_allele = updated_alt_allele
mut.start = updated_start
if ctr == 0:
fieldnames2Render = MutUtils.getAllAttributeNames(mut)
if sampleNameAnnotationName is not None:
fieldnames2Render += [sampleNameAnnotationName]
for fieldname in fieldnames2Render: # fieldnames that start "_" aren't rendered
if fieldname.startswith("_"):
fieldnames2Render.remove(fieldname)
writer = csv.DictWriter(fptr, fieldnames2Render, extrasaction='ignore', delimiter=self.delimiter,
lineterminator=self.lineterminator)
writer.writeheader()
writer.writerow(mut)
ctr += 1
if (ctr % 1000) == 0:
self.logger.info("Wrote " + str(ctr) + " mutations to tsv.")
sampleNames = list(sampleNames)
sampleNames.sort()
chroms = list(chroms)
return chroms, sampleNames, tempTsvFile.name
class OnpQueue(object):
"""
Bookkeeping class to maintain the mutations waiting to be combined
"""
def __init__(self, mutations):
"""
Initialize an new queue with a MutationData iterator
:param mutations: any MutationData producing Iterator
"""
self.mutations = more_itertools.peekable(mutations)
self.sns = SampleNameSelector(self.mutations.peek())
self.queue = collections.defaultdict(list)
self.indel_queue = []
self.last = 0
self.logger = logging.getLogger(__name__)
self.warned_about_order = False
@staticmethod
def _create_start_position_dict(mutations):
"""
Create a start_position -> mutation dict
:param mutations: a collection of MutationData
:return: a dictionary containing all the input MutationData grouped by start postion
"""
assert (mutations is not None)
starts = collections.defaultdict(list)
for mut in mutations:
starts[int(mut.start)] += [mut]
return starts
@staticmethod
def _paths(finished_paths, path_so_far, start, muts):
"""Return all paths from the start position through the mutation graph
:param finished_paths: completed paths
:param path_so_far: the accumulated mutation->mutation path so far
:param start: the start position to travers the muts from
:param muts: a dictionary in the form {start_position: [Mutation]}
:return: All paths through adjacent mutations starting with mutations at chromosome position start
"""
if muts == [] or start not in muts:
finished_paths.append(path_so_far)
else:
# return reduce(operator.concat, lambda mut: OnpCombiner._paths(path + [mut], mut.end+1, muts), [])
# path = map(lambda mut: OnpQueue._paths(path + [mut], int(mut.end)+1, muts), muts[start])
for mut in muts[start]:
OnpQueue._paths(finished_paths, path_so_far + [mut],
int(mut.end) + 1, muts)
return finished_paths
#return reduce(operator.concat, path)
def _add(self, mutation):
variant_type = TranscriptProviderUtils.infer_variant_type(
mutation.ref_allele, mutation.alt_allele)
# only combine ONPs, not indels
if not TranscriptProviderUtils.is_xnp(variant_type):
self.indel_queue.append(mutation)
else:
self.queue[self.sns.getSampleName(mutation)].append(mutation)
def _walk_mutation_paths(self, muts):
"""
Find all paths through adjacent mutations and return those as combined mutations
Find the first mutations by chromosome position and compute all paths through adjacent mutations reachable from them.
If there are any nodes that were not reached, choose the first position with unreached nodes and repeat
:param muts: a list of mutations to walk throught
:return: a list of new mutations combined from
"""
unreached = muts
paths = []
starts = self._create_start_position_dict(muts)
while unreached:
paths += self._paths([], [],
min([int(mut.start) for mut in unreached]),
starts)
reached = [mut for path in paths for mut in path]
unreached = [mut for mut in muts if mut not in reached]
paths = [OnpQueue._combine_mutations(path) for path in paths]
return paths
def _dump_all(self):
results = []
for (sample, muts) in self.queue.iteritems():
results += self._walk_mutation_paths(muts)
#add all stored up indels
results += self.indel_queue or []
self.indel_queue = []
results.sort(key=lambda x: (int(x.start), int(x.end)))
self.queue.clear()
return results
def _get_all_values(self):
return [j for i in self.queue.values() for j in i]
def _is_adjacent_to_any_xnp(self, new_mutation):
return self._is_adjacent(new_mutation, self._get_all_values())
def _is_adjacent(self, new_mutation, mutations):
if mutations:
ends = [int(x.end) for x in mutations]
return int(new_mutation.start) <= 1 + max(ends)
else:
return False
@staticmethod
def _combine_mutations(mutations):
"""
Merge multiple adjacent mutations into a single new mutation.
:param mutations: an ordered list of MutationData
:returns a new MutationData
:warning: _combine_mutations does not make any attempt to sanity check input mutations
it will happily combine overlapping and non-adjacent mutations on disparate chromosomes
"""
if len(mutations) == 0:
return None
if len(mutations) == 1:
return mutations[0]
# special logic for the attributes
start = min([mut.start for mut in mutations])
end = max([mut.end for mut in mutations])
chr = mutations[0].chr
ref = "".join([mut.ref_allele for mut in mutations])
alt = "".join([mut.alt_allele for mut in mutations])
build = "|".join(set([x.build for x in mutations]))
#create the new mutation
newmut = MutationData(chr=chr,
start=start,
end=end,
ref_allele=ref,
alt_allele=alt,
build=build)
#add annotations to the mutation
allAnnotations = set(flatmap(lambda x: x.keys(), mutations))
annotationNames = allAnnotations - set(
mutations[0].getAttributeNames())
for annotName in annotationNames:
annotations = []
for mut in mutations:
try:
annotations.append(mut.getAnnotation(annotName))
except KeyError:
pass
values = sorted(
(set([x.getValue() for x in annotations if x.getValue()])))
value = "|".join(values)
tags = sorted(set(flatmap(lambda x: x.getTags(), annotations)))
source = annotations[0].getDatasource()
datatype = annotations[0].getDataType()
number = annotations[0].getNumber()
description = annotations[0].getDescription()
newmut.createAnnotation(annotationName=annotName,
annotationValue=value,
annotationSource=source,
annotationDataType=datatype,
annotationDescription=description,
tags=tags,
number=number)
return newmut
def _combine_with_indels(self, output):
"""add the indels to output and sort by start position"""
output += self.indel_queue or []
self.indel_queue = []
output.sort(key=lambda x: int(x.start))
return
def get_combined_mutations(self):
"""
:return: a generator yielding mutations, adjacent SNPs,DNPs, and ONPs will be merged together.
"""
# assumes mutations are sorted by start position and then sample
#if they're not, it won't find DNPs
last_chr = -1
last_start = -1
for mut in self.mutations:
output = []
#if we're on a new chromosome, dump all mutations, then add the new one to the queue
if mut.chr != last_chr:
output = self._dump_all()
self._add(mut)
#if we're at the same start position, add the new mutation to the queue
elif mut.start == last_start:
self._add(mut)
#if we are at a new position on the same chromosome
elif self._is_adjacent_to_any_xnp(mut):
# if we are adjacent/overlapping to one of our existing positions
# add the mutation
if not self.warned_about_order and int(mut.start) < last_start:
self.logger.warn(
"Mutations are not sorted by start position, this may cause unexpected behavior or "
"increased memory requirements. It is recommended that your sort any files that you"
"using with --infer-onps by position and sample name.")
self.warned_about_order = True
self._add(mut)
# if we are not adjacent to any existing queue position,
# dump mutations, then add the mutation
else:
output = self._dump_all()
self._add(mut)
last_chr = mut.chr
last_start = mut.start
for mut in output:
yield mut
#when we're finished, be sure to dump any last mutations
output = self._dump_all()
for mut in output:
yield mut
class OnpQueue(object):
"""
Bookkeeping class to maintain the mutations waiting to be combined
"""
def __init__(self, mutations, mutation_data_factory):
"""
Initialize an new queue with a MutationData iterator
:param mutations: any MutationData producing Iterator
:param mutation_data_factory: a MutationDataFactory to be used to produce new mutations for the ONPs
"""
self.mutations = more_itertools.peekable(mutations)
self.sns = SampleNameSelector(self.mutations.peek())
self.queue = collections.defaultdict(list)
self.indel_queue = []
self.last = 0
self.logger = logging.getLogger(__name__)
self.warned_about_order = False
self._mutation_data_factory = mutation_data_factory
@staticmethod
def _create_start_position_dict(mutations):
"""
Create a start_position -> mutation dict
:param mutations: a collection of MutationData
:return: a dictionary containing all the input MutationData grouped by start postion
"""
assert mutations is not None
starts = collections.defaultdict(list)
for mut in mutations:
starts[int(mut.start)] += [mut]
return starts
@staticmethod
def _paths(finished_paths, path_so_far, start, muts):
"""Return all paths from the start position through the mutation graph
:param finished_paths: completed paths
:param path_so_far: the accumulated mutation->mutation path so far
:param start: the start position to travers the muts from
:param muts: a dictionary in the form {start_position: [Mutation]}
:return: All paths through adjacent mutations starting with mutations at chromosome position start
"""
if muts == [] or start not in muts:
# No mutations available to continue this chain
finished_paths.append(path_so_far)
else:
# return reduce(operator.concat, lambda mut: OnpCombiner._paths(path + [mut], mut.end+1, muts), [])
# path = map(lambda mut: OnpQueue._paths(path + [mut], int(mut.end)+1, muts), muts[start])
for mut in muts[start]:
if len(path_so_far) > 0 and not PhasingUtils.is_in_phase(path_so_far[-1], mut):
# Next mutation not in phase, so stop this path here.
finished_paths.append(path_so_far)
else:
OnpQueue._paths(finished_paths, path_so_far + [mut], int(mut.end) + 1, muts)
return finished_paths
# return reduce(operator.concat, path)
def _add(self, mutation):
variant_type = TranscriptProviderUtils.infer_variant_type(mutation.ref_allele, mutation.alt_allele)
# only combine ONPs, not indels
if not TranscriptProviderUtils.is_xnp(variant_type):
self.indel_queue.append(mutation)
else:
self.queue[self.sns.getSampleName(mutation)].append(mutation)
def _walk_mutation_paths(self, muts):
"""
Find all paths through adjacent mutations and return those as combined mutations
Find the first mutations by chromosome position and compute all paths through adjacent mutations reachable from them.
If there are any nodes that were not reached, choose the first position with unreached nodes and repeat
:param muts: a list of mutations to walk throught
:return: a list of new mutations combined from
"""
unreached = muts
paths = []
starts = self._create_start_position_dict(muts)
while unreached:
paths += self._paths([], [], min([int(mut.start) for mut in unreached]), starts)
reached = [mut for path in paths for mut in path]
unreached = [mut for mut in muts if mut not in reached]
paths = [OnpQueue._combine_mutations(path, self._mutation_data_factory) for path in paths]
return paths
def _dump_all(self):
results = []
for (sample, muts) in self.queue.iteritems():
results += self._walk_mutation_paths(muts)
# add all stored up indels
results += self.indel_queue or []
self.indel_queue = []
results.sort(key=lambda x: (int(x.start), int(x.end)))
self.queue.clear()
return results
def _get_all_values(self):
return [j for i in self.queue.values() for j in i]
def _is_adjacent_to_any_xnp(self, new_mutation):
return self._is_adjacent(new_mutation, self._get_all_values())
def _is_adjacent(self, new_mutation, mutations):
if mutations:
ends = [int(x.end) for x in mutations]
return int(new_mutation.start) <= 1 + max(ends)
else:
return False
@staticmethod
def _combine_mutations(mutations, mutation_data_factory):
"""
Merge multiple adjacent mutations into a single new mutation.
:param mutations: an ordered list of MutationData
:returns a new MutationData
:warning: _combine_mutations does not make any attempt to sanity check input mutations
it will happily combine overlapping and non-adjacent mutations on disparate chromosomes
"""
if len(mutations) == 0:
return None
if len(mutations) == 1:
return mutations[0]
# special logic for the attributes
start = min([mut.start for mut in mutations])
end = max([mut.end for mut in mutations])
chr = mutations[0].chr
ref = "".join([mut.ref_allele for mut in mutations])
alt = "".join([mut.alt_allele for mut in mutations])
build = "|".join(set([x.build for x in mutations]))
# create the new mutation
newmut = mutation_data_factory.create(
chr=chr, start=start, end=end, ref_allele=ref, alt_allele=alt, build=build
)
# add annotations to the mutation
allAnnotations = set(flatmap(lambda x: x.keys(), mutations))
annotationNames = allAnnotations - set(mutations[0].getAttributeNames())
for annotName in annotationNames:
annotations = []
for mut in mutations:
try:
annotations.append(mut.getAnnotation(annotName))
except KeyError:
pass
values = [x.getValue() for x in annotations]
if len(set(values)) == 1:
value = values[0] # if all annotations are identical then don't pipe separate them
else:
value = "|".join(values)
tags = sorted(set(flatmap(lambda x: x.getTags(), annotations)))
source = annotations[0].getDatasource()
datatype = annotations[0].getDataType()
number = annotations[0].getNumber()
description = annotations[0].getDescription()
newmut.createAnnotation(
annotationName=annotName,
annotationValue=value,
annotationSource=source,
annotationDataType=datatype,
annotationDescription=description,
tags=tags,
number=number,
)
return newmut
def _combine_with_indels(self, output):
"""add the indels to output and sort by start position"""
output += self.indel_queue or []
self.indel_queue = []
output.sort(key=lambda x: int(x.start))
return
def get_combined_mutations(self):
"""
:return: a generator yielding mutations, adjacent SNPs,DNPs, and ONPs will be merged together.
"""
# assumes mutations are sorted by start position and then sample
# if they're not, it won't find DNPs
last_chr = -1
last_start = -1
for mut in self.mutations:
output = []
# if we're on a new chromosome, dump all mutations, then add the new one to the queue
if mut.chr != last_chr:
output = self._dump_all()
self._add(mut)
# if we're at the same start position, add the new mutation to the queue
elif mut.start == last_start:
self._add(mut)
# if we are at a new position on the same chromosome
elif self._is_adjacent_to_any_xnp(mut):
# if we are adjacent/overlapping to one of our existing positions
# add the mutation
if not self.warned_about_order and int(mut.start) < last_start:
self.logger.warn(
"Mutations are not sorted by start position, this may cause unexpected behavior or "
"increased memory requirements. It is recommended that your sort any files that you"
"using with --infer-onps by position and sample name."
)
self.warned_about_order = True
self._add(mut)
# if we are not adjacent to any existing queue position,
# dump mutations, then add the mutation
else:
output = self._dump_all()
self._add(mut)
last_chr = mut.chr
last_start = mut.start
for mut in output:
yield mut
# when we're finished, be sure to dump any last mutations
output = self._dump_all()
for mut in output:
yield mut
