Triangle Multiplication Is All You Need For Biomolecular Structure Representations

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ouyang-Zhang, Jeffrey, Murugan, Pranav, Diaz, Daniel J., Scarpellini, Gianluca, Bowen, Richard Strong, Gruver, Nate, Klivans, Adam, Krähenbühl, Philipp, Faust, Aleksandra, Al-Shedivat, Maruan
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915652735860736
author Ouyang-Zhang, Jeffrey
Murugan, Pranav
Diaz, Daniel J.
Scarpellini, Gianluca
Bowen, Richard Strong
Gruver, Nate
Klivans, Adam
Krähenbühl, Philipp
Faust, Aleksandra
Al-Shedivat, Maruan
author_facet Ouyang-Zhang, Jeffrey
Murugan, Pranav
Diaz, Daniel J.
Scarpellini, Gianluca
Bowen, Richard Strong
Gruver, Nate
Klivans, Adam
Krähenbühl, Philipp
Faust, Aleksandra
Al-Shedivat, Maruan
contents AlphaFold has transformed protein structure prediction, but emerging applications such as virtual ligand screening, proteome-wide folding, and de novo binder design demand predictions at a massive scale, where runtime and memory costs become prohibitive. A major bottleneck lies in the Pairformer backbone of AlphaFold3-style models, which relies on computationally expensive triangular primitives-especially triangle attention-for pairwise reasoning. We introduce Pairmixer, a streamlined alternative that eliminates triangle attention while preserving higher-order geometric reasoning capabilities that are critical for structure prediction. Pairmixer substantially improves computational efficiency, matching state-of-the-art structure predictors across folding and docking benchmarks, delivering up to 4x faster inference on long sequences while reducing training cost by 34%. Its efficiency alleviates the computational burden of downstream applications such as modeling large protein complexes, high-throughput ligand and binder screening, and hallucination-based design. Within BoltzDesign, for example, Pairmixer delivers over 2x faster sampling and scales to sequences ~30% longer than the memory limits of Pairformer. Code is available at https://github.com/genesistherapeutics/pairmixer.
format Preprint
id arxiv_https___arxiv_org_abs_2510_18870
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Triangle Multiplication Is All You Need For Biomolecular Structure Representations
Ouyang-Zhang, Jeffrey
Murugan, Pranav
Diaz, Daniel J.
Scarpellini, Gianluca
Bowen, Richard Strong
Gruver, Nate
Klivans, Adam
Krähenbühl, Philipp
Faust, Aleksandra
Al-Shedivat, Maruan
Quantitative Methods
Machine Learning
AlphaFold has transformed protein structure prediction, but emerging applications such as virtual ligand screening, proteome-wide folding, and de novo binder design demand predictions at a massive scale, where runtime and memory costs become prohibitive. A major bottleneck lies in the Pairformer backbone of AlphaFold3-style models, which relies on computationally expensive triangular primitives-especially triangle attention-for pairwise reasoning. We introduce Pairmixer, a streamlined alternative that eliminates triangle attention while preserving higher-order geometric reasoning capabilities that are critical for structure prediction. Pairmixer substantially improves computational efficiency, matching state-of-the-art structure predictors across folding and docking benchmarks, delivering up to 4x faster inference on long sequences while reducing training cost by 34%. Its efficiency alleviates the computational burden of downstream applications such as modeling large protein complexes, high-throughput ligand and binder screening, and hallucination-based design. Within BoltzDesign, for example, Pairmixer delivers over 2x faster sampling and scales to sequences ~30% longer than the memory limits of Pairformer. Code is available at https://github.com/genesistherapeutics/pairmixer.
title Triangle Multiplication Is All You Need For Biomolecular Structure Representations
topic Quantitative Methods
Machine Learning
url https://arxiv.org/abs/2510.18870