HMS Research Computing is pleased to announce that Alphafold and LocalColabFold are now officially available (at limited throughput) for research use on O2. This software provides a quantum leap in protein structure prediction, critical for a wide range of life sciences research.

Alphafold is a cutting edge software developed by Google's DeepMind which predicts protein structures from their base amino acid sequences. LocalColabfold is a modified version of Alphafold that uses an independent sequence alignment strategy which allows it to run faster than Alphafold in most use cases.

On June 8, 2022, HMS Research Computing presented a town hall with the Center for Computational Biomedicine (CCB) and protein researchers to introduce the AlphaFold and ColabFold protein structure prediction tools and their research potential. Over 160 people attended this forum discussing usage, the interpretation of results, and applications.

This town hall was part of an ongoing collaboration to make AlphaFold and LocalColabFold accessible to the HMS and affiliated research community, and was driven by unprecedented community enthusiasm and requests to make these tools available. A recording of the event as well as presentation materials are available on the Center for Computational Biomedicine's website.

Research Computing hopes that this event, alongside our documentation on using AlphaFold/LocalColabFold on O2, will enable both experienced researchers and those that have not used O2 before to use this groundbreaking software. The documentation provides step-by-step instructions for accessing and executing these tools in the HPC environment, including template scripts, and describes best practices for running these jobs on the cluster.

Research Computing will continue to improve support for these tools based on community feedback and internal testing as well as continued collaboration with CCB, and will aim to further polish the experience for the user community as development on these tools continues.