OkayBoboLi's picture
Add the PROTAC release
f12d3b3 verified
Raw History Blame Contribute Delete
2.04 kB
CupOFMoCA — PROTAC release
Copyright 2026 CupOFMoCA authors
The first-party code in this release is licensed under Apache License 2.0; see
LICENSE. Model weights and third-party downloads are subject to their own terms,
which users are responsible for reviewing.
Redistributed third-party code
------------------------------
src/defog/
DeFoG, https://github.com/manuelmlmadeira/DeFoG
MIT License, reproduced at src/defog/LICENSE.
src/defog/src/datasets/protac_dataset.py is a first-party addition.
The upstream graph-benchmark evaluation helpers under
src/defog/src/analysis/orca/ are NOT redistributed: that C++ code carries no
license header and is unused by the PROTAC pipeline.
Third-party code that is NOT redistributed
------------------------------------------
Two optional components are published by their authors without any license
file, so this release cannot lawfully redistribute them. Run
`python scripts/fetch_third_party.py` to obtain them from their upstream
repositories at the commits used for the paper, and review the upstream terms
before use.
DegradeMaster https://github.com/ABILiLab/DegradeMaster
pinned commit aa149beaf067a051e070b3b281f7dba43c2f3e90
Needed only for predicted-degradation scoring.
EGFR-PROPK https://github.com/Zhang-Ran-0119/EGFR-PROPK
pinned commit 6940cb90c449e6c6d0678508fc2f228da796e028
Needed only for the EGFR-only PK columns of batch_report.py.
Linker generation with scripts/generate.py does not require either component.
Derived data
------------
The PROTAC-DB derived tables (data/linker.csv) and the target and E3 pocket
graphs used for degradation scoring were prepared by the CupOFMoCA authors from
publicly available sources: PROTAC-DB (Weng et al., Nucleic Acids Research 2021)
and the RCSB Protein Data Bank. The fine-tuned generator checkpoint under
checkpoints/ is a derivative of the released DeFoG GuacaMol checkpoint and
inherits its MIT terms.