def add_quality(hmmcopy_metrics, alignment_metrics, multipliers, output, training_data, tempdir):
helpers.makedirs(tempdir)
hmmcopy_tables = ['/hmmcopy/metrics/{}'.format(mult) for mult in multipliers]
model = classify.train_classifier(training_data)
feature_names = model.feature_names_
data = classify.load_data(hmmcopy_metrics, alignment_metrics,
hmmcopy_tables, '/alignment/metrics',
feature_names)
for i, (hmmcopy_table, tabledata) in enumerate(data):
intermediate_output = os.path.join(
tempdir, '{}_metrics_with_quality.csv.gz'.format(i)
)
predictions = classify.classify(model, tabledata)
classify.write_to_output(
hmmcopy_metrics,
hmmcopy_table,
intermediate_output,
predictions)
csvutils.prep_csv_files(intermediate_output, output, dtypes=dtypes()['metrics'])
def add_quality(hmmcopy_metrics, alignment_metrics, output, training_data,
tempdir):
helpers.makedirs(tempdir)
intermediate_output = os.path.join(tempdir, 'metrics_with_quality.csv')
model = classify.train_classifier(training_data)
feature_names = model.feature_names_
data = classify.load_data(hmmcopy_metrics, alignment_metrics,
feature_names)
predictions = classify.classify(model, data)
classify.write_to_output(hmmcopy_metrics, intermediate_output, predictions)
csvutils.prep_csv_files(intermediate_output, output)
def run_hmmcopy(
bam_file,
corrected_reads_filename,
segments_filename,
parameters_filename,
metrics_filename,
hmmcopy_tar,
cell_id,
hmmparams,
tempdir,
docker_image
):
# generate wig file for hmmcopy
helpers.makedirs(tempdir)
readcount_wig = os.path.join(tempdir, 'readcounter.wig')
corrected_reads = os.path.join(tempdir, 'corrected_reads.csv')
run_correction_hmmcopy(
bam_file,
corrected_reads,
readcount_wig,
hmmparams,
docker_image
)
hmmcopy_tempdir = os.path.join(tempdir, '{}_hmmcopy'.format(cell_id))
helpers.makedirs(hmmcopy_tempdir)
run_hmmcopy_script(
corrected_reads,
hmmcopy_tempdir,
cell_id,
hmmparams,
docker_image
)
hmmcopy_outdir = os.path.join(hmmcopy_tempdir, str(0))
csvutils.prep_csv_files(
os.path.join(hmmcopy_outdir, "reads.csv"), corrected_reads_filename
)
csvutils.prep_csv_files(
os.path.join(hmmcopy_outdir, "params.csv"), parameters_filename
)
csvutils.prep_csv_files(
os.path.join(hmmcopy_outdir, "segs.csv"), segments_filename
)
csvutils.prep_csv_files(
os.path.join(hmmcopy_outdir, "metrics.csv"), metrics_filename
)
helpers.make_tarfile(hmmcopy_tar, hmmcopy_tempdir)