def create_ref_id_dict(self, ref_genomes):
mydict = {}
for ref in ref_genomes:
mydict[misc.get_refid_from_path(ref)] = ref
if len(mydict) != len(ref_genomes):
raise TypeError("The given reference genomes must have unique base names in their paths.")
return mydict
def align_to_references(state):
logger = state.logger
ref_sams = {}
for refpath in state.paths.reference_genomes:
suffix = basename(refpath).split('.')[0]
logger.info("Building the genome fasta file for the file %s..." %
basename(refpath))
bowtie2.build('2.2.9', refpath, logger)
logger.info("Aligning the forward reads in %s to the genome %s..." %
(basename(state.paths.fastq_files[0]), basename(refpath)))
outsam1 = bowtie2.align_to_genome(
'2.2.9', refpath, state.paths.fastq_files[0],
os.path.join(state.paths.full_sams_dir,
'fastq1_to_genome_%s.sam' % suffix),
state.settings.threads)
logger.info("Output saved to %s" % basename(outsam1))
logger.debug("Saving reference header for %s under %s" %
(basename(refpath), misc.get_refid_from_path(refpath)))
state.settings.reference_headers_dict[misc.get_refid_from_path(
refpath)].append(misc.get_sam_header(outsam1))
logger.info("Aligning the reverse reads in %s to the genome %s..." %
(basename(state.paths.fastq_files[1]), basename(refpath)))
outsam2 = bowtie2.align_to_genome(
'2.2.9', refpath, state.paths.fastq_files[1],
os.path.join(state.paths.full_sams_dir,
'fastq2_to_genome_%s.sam' % suffix),
state.settings.threads)
logger.info("Output saved to %s" % basename(outsam2))
logger.debug("Saving reference header for %s under %s" %
(basename(refpath), misc.get_refid_from_path(refpath)))
state.settings.reference_headers_dict[misc.get_refid_from_path(
refpath)].append(misc.get_sam_header(outsam2))
ref_sams[refpath] = (outsam1, outsam2)
state.paths.fastq_to_genome_algnmnts = ref_sams
return ref_sams
def genome(path):
global ref_basenames
try:
if not os.path.isfile(path):
raise TypeError("Genome fasta is not a file.")
if misc.get_refid_from_path(path) in ref_basenames:
raise TypeError("Duplicate Basename")
ref_basenames.add(os.path.basename(path).split('.')[0])
return os.path.abspath(path)
except:
raise argparse.ArgumentTypeError(
"Genomes must be entered as \"<reference-fasta1> <reference-fasta2>\" and "
"they must have unique basenames.")