Efficient protein structure prediction from compact computers to datacenters with OpenFold-TRT
Preprint
Abstract
We introduce accelerations for deep learning inference with OpenFold and TensorRT that, combined with MMseqs2-GPU on an x86 system with one NVIDIA RTX PRO 6000 Blackwell Server Edition GPU, reach up to 131× faster inference compared to AlphaFold2. ARM-optimizations enable homology search on power efficient DGX Spark, and allow search beyond available GPU RAM on NVIDIA Grace Hopper Superchip at comparable overall folding speed. These accelerations enable high-throughput protein structure inference at no accuracy cost using familiar tools.Keywords
This publication has 20 references indexed in Scilit:
- GPU-accelerated homology search with MMseqs2Nature Methods, 2025
- OpenFold: retraining AlphaFold2 yields new insights into its learning mechanisms and capacity for generalizationNature Methods, 2024
- Accurate structure prediction of biomolecular interactions with AlphaFold 3Nature, 2024
- ColabFold: making protein folding accessible to allNature Methods, 2022
- Critical assessment of methods of protein structure prediction (CASP)—Round XIVProteins: Structure, Function, and Bioinformatics, 2021
- Highly accurate protein structure prediction with AlphaFoldNature, 2021
- Clustering huge protein sequence sets in linear timeNature Communications, 2018
- MMseqs2 enables sensitive protein sequence searching for the analysis of massive data setsNature Biotechnology, 2017
- Three-Dimensional Structures of Membrane Proteins from Genomic SequencingCell, 2012
- Direct-coupling analysis of residue coevolution captures native contacts across many protein familiesProceedings of the National Academy of Sciences, 2011