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

[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.