About

ParaFold is a modified version of AlphaFold for high performance computing, developed by Shanghai Jiao Tong Univerisity.

AlphaFold uncovers the building blocks of life and paves the way for a new era in science. It is going to be exciting to see the many ways in which it is going to radically change biological investigation. Immediate applications are probably less interesting than what AlphaFold might become the foundation for over next couple decades.

However, as mentioned in the second Nature paper from DeepMind, the AlphaFold2 code open-sourced in July is not the proteome-scale pipeline used for large-scale structure predictions. We need to tune the code for parallel computer systems especially for high performance computing.

So here we are, a team from Shanghai Jiao Tong University to optimize AlphaFold on high performance computing platform, making it faster, more accurate, and more convenient. Optimization is performed at four levels: CPU and GPU separation, CPU parallel optimization, GPU optimization, and I/O optimization.

Let us know what you think and please feel free to reach out if you have any questions or feedback. Open an issue on GitHub or contact us by email zbztzhz[AT]sjtu.edu.cn.

Thanks for reading!