def create():
"""Generates SidechainNet for a single CASP thinning."""
# First, parse raw proteinnet files into Python dictionaries for convenience
pnids = parse_raw_proteinnet(args.proteinnet_in, args.proteinnet_out,
args.training_set)
pnids = pnids[:args.limit] # Limit the length of the list for debugging
# Using the ProteinNet IDs as a guide, download the relevant sidechain data
sc_only_data, sc_filename = download_sidechain_data(
pnids, args.sidechainnet_out, args.casp_version, args.training_set,
args.limit, args.proteinnet_in, args.regenerate_scdata)
# Finally, unify the sidechain data with ProteinNet
sidechainnet_raw = combine_datasets(args.proteinnet_out, sc_only_data,
args.training_set)
sidechainnet_outfile = os.path.join(
args.sidechainnet_out,
format_sidechainnet_path(args.casp_version, args.training_set))
sidechainnet = organize_data(sidechainnet_raw, args.proteinnet_out,
args.casp_version, args.training_set)
save_data(sidechainnet, sidechainnet_outfile)
print(
f"SidechainNet for CASP {args.casp_version} written to {sidechainnet_outfile}."
)
def create_all():
"""Generates all thinnings of a particular CASP dataset, starting with the largest."""
# First, parse raw proteinnet files into Python dictionaries for convenience
pnids = parse_raw_proteinnet(args.proteinnet_in, args.proteinnet_out, 100)
pnids = pnids[:args.limit] # Limit the length of the list for debugging
# Using the ProteinNet IDs as a guide, download the relevant sidechain data
sc_only_data, sc_filename = download_sidechain_data(
pnids,
args.sidechainnet_out,
args.casp_version,
100,
args.limit,
args.proteinnet_in,
regenerate_scdata=args.regenerate_scdata)
# Finally, unify the sidechain data with ProteinNet
sidechainnet_raw_100 = combine_datasets(args.proteinnet_out, sc_only_data, 100)
for training_set in [100, 95, 90, 70, 50, 30]:
sc_outfile = os.path.join(
args.sidechainnet_out,
format_sidechainnet_path(args.casp_version, training_set))
sidechainnet = organize_data(sidechainnet_raw_100, args.proteinnet_out,
args.casp_version, training_set)
save_data(sidechainnet, sc_outfile)
print(f"SidechainNet for CASP {args.casp_version} "
f"({training_set}% thinning) written to {sc_outfile}.")