def summarize_amp_analysis( input_dir, output_dir ):
create_directory( output_dir )
log.info('Combining AmpliconAnalysis output from "{0}" to "{1}"'.format(input_dir, output_dir))
results = list( find_amp_assem_results(input_dir) )
output_files = list( output_amp_assem_results(results, output_dir) )
result_output = os.path.join( output_dir, "AmpliconAssembly_Results.fofn" )
write_list_file( output_files, result_output )
return result_output
def summarize_amp_analysis(input_dir, output_dir):
create_directory(output_dir)
log.info('Combining AmpliconAnalysis output from "{0}" to "{1}"'.format(
input_dir, output_dir))
results = list(find_amp_assem_results(input_dir))
output_files = list(output_amp_assem_results(results, output_dir))
result_output = os.path.join(output_dir, "AmpliconAssembly_Results.fofn")
write_list_file(output_files, result_output)
return result_output
def __init__(self, output, setup=None, nproc=1, debug=False):
"""Initialize cross-cluster and object-specific settings"""
if debug:
log.setLevel(logging.DEBUG)
log.debug("TESTING")
log.info("Initializing Resequencer sub-module")
self._resequencer = Resequencer(setup, nproc)
# Initialize output folder
self._output = output
create_directory(self.output)
def __init__(self, output, setup=None, nproc=1, debug=False):
"""Initialize cross-cluster and object-specific settings"""
if debug:
log.setLevel(logging.DEBUG)
log.debug("TESTING")
log.info("Initializing Resequencer sub-module")
self._resequencer = Resequencer(setup, nproc)
# Initialize output folder
self._output = output
create_directory(self.output)
def combine_clusense_output(input_dir, output_dir):
create_directory( output_dir )
log.info('Combining clusense output from "{0}" in "{1}"'.format(input_dir, output_dir))
clusense_dirs = find_clusense_dirs( input_dir )
clusense_clusters = find_clusense_clusters( clusense_dirs )
cns_files, read_files = output_clusters( clusense_clusters, output_dir )
cns_output = os.path.join( output_dir, CNS_FOFN )
write_list_file( cns_files, cns_output )
read_output = os.path.join( output_dir, READ_FOFN )
write_list_file( read_files, read_output )
return cns_output, read_output
def combine_clusense_output(input_dir, output_dir):
create_directory(output_dir)
log.info('Combining clusense output from "{0}" in "{1}"'.format(
input_dir, output_dir))
clusense_dirs = find_clusense_dirs(input_dir)
clusense_clusters = find_clusense_clusters(clusense_dirs)
cns_files, read_files = output_clusters(clusense_clusters, output_dir)
cns_output = os.path.join(output_dir, CNS_FOFN)
write_list_file(cns_files, cns_output)
read_output = os.path.join(output_dir, READ_FOFN)
write_list_file(read_files, read_output)
return cns_output, read_output
def split_results(amp_analysis):
"""Split the output of an Amplicon Analysis job by Barcode"""
assert os.path.isdir(amp_analysis)
sequence_path = os.path.join(amp_analysis, 'amplicon_analysis.fasta')
check_output_file(sequence_path)
print "Analyzing %s output sequences" % fasta_size(sequence_path)
barcode_path = os.path.join(amp_analysis, 'by_barcode')
create_directory(barcode_path)
records = list(FastaReader(sequence_path))
barcodes = {get_barcode(r): [] for r in records}
[barcodes[get_barcode(r)].append(r) for r in records]
barcode_files = {}
for barcode, records in barcodes.iteritems():
barcode_file = barcode + '.fasta'
sample_path = os.path.join(barcode_path, barcode_file)
with FastaWriter(sample_path) as handle:
for record in records:
handle.writeRecord(record)
barcode_files[barcode] = sample_path
return barcode_files
def split_results(amp_analysis):
"""Split the output of an Amplicon Analysis job by Barcode"""
assert os.path.isdir(amp_analysis)
sequence_path = os.path.join(amp_analysis, "amplicon_analysis.fasta")
check_output_file(sequence_path)
print "Analyzing %s output sequences" % fasta_size(sequence_path)
barcode_path = os.path.join(amp_analysis, "by_barcode")
create_directory(barcode_path)
records = list(FastaReader(sequence_path))
barcodes = {get_barcode(r): [] for r in records}
[barcodes[get_barcode(r)].append(r) for r in records]
barcode_files = {}
for barcode, records in barcodes.iteritems():
barcode_file = barcode + ".fasta"
sample_path = os.path.join(barcode_path, barcode_file)
with FastaWriter(sample_path) as handle:
for record in records:
handle.writeRecord(record)
barcode_files[barcode] = sample_path
return barcode_files
def extract_cDNA( input_file, exon_fofn, output=None,
directory=None,
reference_file=None,
alignment_file=None,
debug=False ):
"""
Extract the cDNA sequences from a mixed Fasta
"""
# Check the input files, and align the input file if needed
if reference_file and alignment_file is None:
alignment_file = align_best_reference( input_file, reference_file )
elif reference_file is None and alignment_file is None:
msg = "extract_alleles requires either an Alignment or a Reference!"
log.error( msg )
raise IOError( msg )
# Set the output and directory if it hasn't been specified
if directory is None:
dirname = os.path.dirname( input_file )
directory = os.path.join( dirname, 'cDNA' )
remove_directory( directory )
create_directory( directory )
output = output or _get_output_file( input_file )
# Prepare the Fasta by orienting and subsetting it
records = _parse_input_records( input_file )
fofn = _parse_exon_fofn( exon_fofn )
loci = _parse_loci( alignment_file )
log.info("Extracting cDNA sequences from all records")
_extract_cDNA( records, loci, fofn, directory )
log.info("Collecting all extracted cDNA records into %s" % output)
_collect_cDNA( directory, output )
# Clean up the directory and return the combined cDNA file
if not debug:
remove_directory( directory )
return output
def initialize_output(self, output):
"""Initialize the cluster-specific output folders"""
# TODO: Check for existing directories and do something
self._output = os.path.abspath(output)
create_directory(self._output)
self._scripts = os.path.join(self._output, "scripts")
create_directory(self._scripts)
self._logs = os.path.join(self._output, "logs")
create_directory(self._logs)
def initialize_output(self, output):
"""Initialize the cluster-specific output folders"""
# TODO: Check for existing directories and do something
self._output = os.path.abspath(output)
create_directory(self._output)
self._scripts = os.path.join(self._output, 'scripts')
create_directory(self._scripts)
self._logs = os.path.join(self._output, 'logs')
create_directory(self._logs)
def _extract_cDNA( records, loci, fofn, directory ):
"""
Extract and create a cDNA record for each Fasta Sequence
"""
for record in records:
# Create an output folder for each record to process
name = record.name.split()[0]
try:
locus = loci[name]
except:
log.warn( 'No HLA locus associated with "%s" - skipping' % name )
continue
# Create a directory
record_directory = os.path.join( directory, name )
create_directory( record_directory )
# Find the appropriate Locus and FOFN
if locus in fofn:
exon_fofn = fofn[locus]
else:
log.warn( 'No exonic reference for %s' % locus )
# Extract the exons and make the cDNA
exon_file = extract_exons( record, exon_fofn, record_directory )
if exon_file:
cDNA_file = exons_to_cDNA( exon_file )
def get_output_folder(self, barcode):
output_dir = os.path.join(self.output, barcode)
create_directory(output_dir)
return output_dir
def get_output_folder(self, barcode):
output_dir = os.path.join(self.output, barcode)
create_directory(output_dir)
return output_dir