@@ -276,6 +276,18 @@ def cmd_compute_fp_retro_compound(args: argparse.Namespace) -> None:
276276 console .print (f"Computed RetroMol fingerprints for { done } entries (batch={ args .batch } )" )
277277
278278
279+ def cmd_compute_fp_retro_gbk (args : argparse .Namespace ) -> None :
280+ """
281+ Compute biosynthetic fingerprints for GenBank records.
282+
283+ :param args: command-line arguments
284+ """
285+ from bionexus .etl .retromol import backfill_retro_fingerprints_gbk
286+
287+ done = backfill_retro_fingerprints_gbk (cache_dir = args .cache_dir , batch = args .batch , recompute = args .recompute , paras_model_path = args .paras )
288+ console .print (f"Computed RetroMol fingerprints for { done } GenBank records (batch={ args .batch } )" )
289+
290+
279291def cmd_dump_db (args : argparse .Namespace ) -> None :
280292 """
281293 Dump the database to a pg_dump custom format file.
@@ -546,6 +558,13 @@ def build_parser() -> argparse.ArgumentParser:
546558 p_do .add_argument ("--recompute" , action = "store_true" , help = "Force recomputation" )
547559 p_do .set_defaults (func = cmd_compute_fp_retro_compound )
548560
561+ p_gbk = sub .add_parser ("compute-fp-retro-gbk" ,help = "Compute biosynthetic fingerprints for GenBank records" )
562+ p_gbk .add_argument ("--cache-dir" , default = None , help = "Cache/work dir for RetroMol" )
563+ p_gbk .add_argument ("--batch" , type = int , default = 2000 )
564+ p_gbk .add_argument ("--recompute" , action = "store_true" , help = "Force recomputation" )
565+ p_gbk .add_argument ("--paras" , required = True , help = "Path to the PARAS model file" )
566+ p_gbk .set_defaults (func = cmd_compute_fp_retro_gbk )
567+
549568 p_dump = sub .add_parser ("dump-db" , help = "Write pg_dump custom format" )
550569 p_dump .add_argument ("--out" , default = "dumps/bionexus.dump" )
551570 p_dump .set_defaults (func = cmd_dump_db )
0 commit comments