HelixFold: An Efficient Implementation of AlphaFold2 using PaddlePaddle

TL;DR

HelixFold is an optimized implementation of AlphaFold2 using PaddlePaddle that significantly reduces training time and memory usage while maintaining high accuracy, making protein structure prediction more accessible.

Contribution

This work introduces HelixFold, a PaddlePaddle-based implementation of AlphaFold2 that improves efficiency and reduces resource requirements compared to existing versions.

Findings

HelixFold trains in 7.5 days, faster than 11 days for AlphaFold2 and OpenFold.

HelixFold maintains comparable accuracy to AlphaFold2 on CASP14 and CAMEO datasets.

Memory consumption is optimized through various techniques, enabling more accessible protein structure prediction.

Abstract

Accurate protein structure prediction can significantly accelerate the development of life science. The accuracy of AlphaFold2, a frontier end-to-end structure prediction system, is already close to that of the experimental determination techniques. Due to the complex model architecture and large memory consumption, it requires lots of computational resources and time to implement the training and inference of AlphaFold2 from scratch. The cost of running the original AlphaFold2 is expensive for most individuals and institutions. Therefore, reducing this cost could accelerate the development of life science. We implement AlphaFold2 using PaddlePaddle, namely HelixFold, to improve training and inference speed and reduce memory consumption. The performance is improved by operator fusion, tensor fusion, and hybrid parallelism computation, while the memory is optimized through Recompute, BFloat16, and memory read/write in-place. Compared with the original AlphaFold2 (implemented with Jax) and OpenFold (implemented with PyTorch), HelixFold needs only 7.5 days to complete the full end-to-end training and only 5.3 days when using hybrid parallelism, while both AlphaFold2 and OpenFold take about 11 days. HelixFold saves 1x training time. We verified that HelixFold's accuracy could be on par with AlphaFold2 on the CASP14 and CAMEO datasets. HelixFold's code is available on GitHub for free download: https://github.com/PaddlePaddle/PaddleHelix/tree/dev/apps/protein_folding/helixfold, and we also provide stable web services on https://paddlehelix.baidu.com/app/drug/protein/forecast.