FKSFold-Chai

FKSFold applies Feynman-Kac (FK) steering to guide the diffusion process in AlphaFold3-type models for molecular glue induced ternary structure prediction. This repository contains the implementation of our early approach that we explored before developing YDS-GlueFold, our more comprehensive and successful model for predicting molecular glue ternary complexes.

Installation

pip install git+https://github.com/YDS-Pharmatech/FKSFold-Chai.git

Usage

Our usage is mostly compatable with the Chai-1 repo. We removed the num_diffn_samples parameter in the run_inference function. You can reference to Chai-1's README for more details.

Command line inference

You can fold a FASTA file containing all the sequences (including modified residues, nucleotides, and ligands as SMILES strings) in a complex of interest by calling:

fksfold fold input.fasta output_folder

Pythonic inference

python examples/run_save_traj.py

Citation

The FKSFold repo is highly relied on the Chai-1 repo. If you found this repo useful, please cite the following:

@article{FKSFold-Technical-Report,
	title        = {FKSFold: Improving AlphaFold3-Type Predictions of Molecular Glue-Induced Ternary Complexes with Feynman-Kac-Steered Diffusion},
	author       = {Shen, Jian and Zhou, Shengmin and Che, Xing},
	year         = 2025,
	journal      = {bioRxiv},
	publisher    = {Cold Spring Harbor Laboratory},
	doi          = {10.1101/2025.05.03.651455},
	url          = {https://www.biorxiv.org/content/10.1101/2025.05.03.651455v1},
	elocation-id = {2025.05.03.651455},
	eprint       = {https://www.biorxiv.org/content/10.1101/2025.05.03.651455v1.full.pdf}
}

@article{Chai-1-Technical-Report,
	title        = {Chai-1: Decoding the molecular interactions of life},
	author       = {{Chai Discovery}},
	year         = 2024,
	journal      = {bioRxiv},
	publisher    = {Cold Spring Harbor Laboratory},
	doi          = {10.1101/2024.10.10.615955},
	url          = {https://www.biorxiv.org/content/early/2024/10/11/2024.10.10.615955},
	elocation-id = {2024.10.10.615955},
	eprint       = {https://www.biorxiv.org/content/early/2024/10/11/2024.10.10.615955.full.pdf}
}

Additionally, if you use the automatic MMseqs2 MSA generation described above, please also cite:

@article{mirdita2022colabfold,
  title={ColabFold: making protein folding accessible to all},
  author={Mirdita, Milot and Sch{\"u}tze, Konstantin and Moriwaki, Yoshitaka and Heo, Lim and Ovchinnikov, Sergey and Steinegger, Martin},
  journal={Nature methods},
  year={2022},
}