def assign_read_kmers(params):
"""
Assigns a single read to a cell barcode by kmer compatibility
args (tuple)
kmers_to_paths: dict of kmer -> list of paths that contain it
min_kmer_size
max_kmer_size
read: list of fastq entry lines
"""
(kmer_map,
min_kmer_size,
max_kmer_size,
(reads_data, reads_offset),
(barcodes_data, barcodes_offset)) = params
for kmer_size in range(max_kmer_size, min_kmer_size, -1):
read_kmers = IO_utils.get_cyclic_kmers(
barcodes_data,
kmer_size,
args['barcode_start'],
args['barcode_end'],
indel = True)
bcs, is_assigned, is_unique = get_most_common_bc(
kmer_map, read_kmers)
if is_assigned and is_unique:
return (bcs[0], reads_offset, barcodes_offset)
#outherwise decrement kmer size and try again
return ('unassigned', reads_offset, barcodes_offset)
def build_subgraph(reads_in_subgraph, barcodes_unzipped): bc_file = open(barcodes_unzipped, 'rb') barcodes_iter = IO_utils.read_fastq_random( bc_file, offsets = reads_in_subgraph) subgraph_kmer_counts = Counter() while(True): try: barcode_data, _ = next(barcodes_iter) except StopIteration: break read_kmers = IO_utils.get_cyclic_kmers( barcode_data, int(args['kmer_size']), int(args['barcode_start']), int(args['barcode_end'])) for (kmer, _ ) in read_kmers: subgraph_kmer_counts[kmer] += 1 bc_file.close() edges = [] for(kmer, count) in subgraph_kmer_counts.items(): edge = Edge(kmer[0:-1], kmer[1:], count) edges.append(edge) subgraph = Graph(edges) return subgraph
def map_kmers_to_bcs_fixed_k(consensus_bcs, kmer_size):
kmers_to_paths = {}
for cell_barcode in consensus_bcs:
kmers = IO_utils.get_cyclic_kmers(
['na', cell_barcode, 'na', cell_barcode],
kmer_size,
0,
len(cell_barcode),
indel=True)
for (kmer, _) in kmers:
if (kmer not in kmers_to_paths.keys()):
kmers_to_paths[kmer] = []
kmers_to_paths[kmer].append(cell_barcode)
return kmers_to_paths
def index_read(params):
"""
Args
params (tuple):
barcodes_data (str): sequence of read_1 (barcode)
barcodes_offset (int): line offset for this read
Returns
kmer_index (dict):
"""
(barcodes_data, barcodes_offset) = params
kmer_index = {}
read_kmers = IO_utils.get_cyclic_kmers(barcodes_data, args['kmer_size'],
args['barcode_start'],
args['barcode_end'])
for (kmer, _) in read_kmers:
if (kmer not in kmer_index.keys()):
kmer_index[kmer] = []
kmer_index[kmer].append(barcodes_offset)
return kmer_index