arxiv:2507.02025

IntFold: A Controllable Foundation Model for General and Specialized Biomolecular Structure Prediction

Published on Jul 2

· Submitted by

siqisun on Jul 4

Upvote

Authors:

Gary Liu ,

Abstract

IntFold uses a customized attention kernel for biomolecular structure prediction, surpassing AlphaFold3, and includes adapters and a novel confidence head for specialized predictions and docking assessments.

AI-generated summary

We introduce IntFold, a controllable foundation model for both general and specialized biomolecular structure prediction. IntFold demonstrates predictive accuracy comparable to the state-of-the-art AlphaFold3, while utilizing a superior customized attention kernel. Beyond standard structure prediction, IntFold can be adapted to predict allosteric states, constrained structures, and binding affinity through the use of individual adapters. Furthermore, we introduce a novel confidence head to estimate docking quality, offering a more nuanced assessment for challenging targets such as antibody-antigen complexes. Finally, we share insights gained during the training process of this computationally intensive model.

View arXiv page View PDF Project page GitHub 53 Add to collection

Community

siqisun

Paper submitter 1 day ago

•

edited 1 day ago

We introduce IntFold, a controllable foundation model for general and specialized biomolecular structure prediction.
codebase: https://github.com/IntelliGen-AI/IntFold
server: https://server.intfold.com/

xavier-robin

about 21 hours ago

•

edited about 20 hours ago

Did you realize that IntFold is already the name of a well-established protein structure prediction server at https://www.reading.ac.uk/bioinf/IntFOLD/?
(Disclaimer: I am not affiliated with IntFOLD).

FuxuLiu

Paper author about 20 hours ago

•

edited about 20 hours ago

Thank you for reaching out regarding the similarity between our server names. We appreciate the important work of IntFOLD Server in the structural bioinformatics community and value your feedback.

We’d like to clarify that our server’s name, IntFold, was derived exclusively from our organization’s identity, Intelligen AI (Int from "Intelligen" + Fold for "structural folding"). At the time of development and launch, we were unaware of the existence of IntFOLD server, and any similarity is purely coincidental.