def preprocessing(fileobj: gzip.GzipFile) -> None:
"""Remove headers and columns to prepare TSV files in mtx directory to make them
compatible for use with ScanPy methods."""
f = pd.read_table(fileobj, sep='\t') # Pandas dataframe
if fileobj.name == 'genes.tsv.gz':
col_to_keep = ['featurekey', 'featurename']
assert col_to_keep[0] in f.columns
assert col_to_keep[1] in f.columns
elif fileobj.name == 'cells.tsv.gz':
fileobj.name = 'barcodes.tsv.gz'
col_to_keep = 'cellkey'
assert col_to_keep in f.columns
else:
raise ValueError(
'Expected genes.tsv.gz and cells.tsv.gz in directory.')
f_new = f[col_to_keep]
# Write to file without column or row headers.
f_new.to_csv(first(os.path.splitext(fileobj.name)),
index=False,
header=False,
sep='\t')
