Skip to content

[Proteina-Complexa] target_schema.md field documentation is inaccurate #46

Description

@xinyu-dev

Several target_schema.md field descriptions have drifted from the code:

  • target_input format — doc lists A1-115/0 50-100 ("RFdiffusion contigs"), but the parser only splits on commas (binder_eval_utils.py:272); only A1-115 and "A96-174,A306-446" work.
  • ligand list vs SMILES — a ligand list is valid for generation (gen_dataset.py:640,669) but requires use_bonds_from_file:True and no SMILES; the add CLI is single-only. The doc doesn't state this exclusivity.
  • ligand_only — doc says "no protein interface"; really True = the whole file is the ligand, False = extract named residue(s) (gen_dataset.py:620). Shipped *_ligand_centered.pdb are ligand-only.

Severity: Medium · Status: doc open (notebooks corrected)

Steps to reproduce

  1. Follow target_schema.md for any field above.
  2. → the slash target_input is silently ignored; a ligand list + SMILES errors; ligand_only behaves opposite to the description.

Fix

Correct the three field descriptions (drop the slash row; document list ⇒ use_bonds_from_file + no SMILES; rewrite ligand_only).

Environment

NVIDIA A100 80GB PCIe · driver 565.57.01 · CUDA 12.7 · repo branch dev @ 916eaae · UV runtime.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions