def read_sequences(filetype, file_list):
if filetype == 'fastq':
return read_fastq_dict(file_list)
elif filetype == 'fasta':
return read_fasta_dict(file_list)
else:
raise ValueError
def read_sequences( filetype, file_list ):
if filetype == 'fastq':
return read_fastq_dict( file_list )
elif filetype == 'fasta':
return read_fasta_dict( file_list )
else:
raise ValueError
def from_assembly( contig_file, reference_fofn ):
contigs = read_fasta_dict( contig_file )
references = read_reference_fofn( reference_fofn )
all_genes = []
for locus, reference in references.iteritems():
alignment = align_reference_to_contigs( locus, reference, contig_file )
hits = read_blasr_hits( alignment )
hits = pick_blasr_hits( hits )
genes = extract_genes( contigs, hits )
all_genes += genes
write_fasta( all_genes, 'output.fasta' )
