Calculate log2 coverages in the BAM file at each interval.
logging.info("Processing reads in %s", os.path.basename(bam_fname))
if procs == 1:
return bedcov(bed_fname, bam_fname, min_mapq)
else:
chunks = []
with futures.ProcessPoolExecutor(procs) as pool:
args_iter = ((bed_chunk, bam_fname, min_mapq)
After Change
for bed_chunk_fname, table in pool.map(_bedcov, args_iter):
chunks.append(table)
rm(bed_chunk_fname)
table = pd.concat(chunks, ignore_index=True)
// Fill in CNA required columns
if "gene" in table:
table["gene"] = table["gene"].fillna("-")
else:
table["gene"] = "-"
// NB: User-supplied bins might be zero-width or reversed -- skip those
spans = table.end - table.start
ok_idx = (spans > 0)
table = table.assign(depth=0, log2=NULL_LOG2_COVERAGE)
table.loc[ok_idx, "depth"] = (table.loc[ok_idx, "basecount"]
/ spans[ok_idx])
ok_idx = (table["depth"] > 0)
table.loc[ok_idx, "log2"] = np.log2(table.loc[ok_idx, "depth"])
return table
def _bedcov(args):
