The next chapter for AI protein structure prediction

When Alphabet announced in 2021 that its AlphaFold AI could predict the structure of almost every known protein, it was a major story, not just in the pharma world, but beyond. Yet, that was far from the end of the story for AI and protein folding.

On today’s pharmaphorum podcast, Glen Gowers, co-founder of Basecamp Research, gets into the company’s recent pre-print paper and how it improves upon AlphaFold 2, the current gold standard for protein folding prediction.

In essence, what Basecamp Research has shown is that replacing the flawed public data sets on which AlphaFold is built with deeper, more accurate, and more representative datasets improves that model’s predictive accuracy sixfold.

Gowers tells host Jonah Comstock about their data set, how it was obtained, and how it improves on the public databases. He also gets into how that data improves the model and looks ahead to how that improved model will affect the growing infrastructure of AI drug discovery based on proteomics.

They also discuss the company’s partnership with NVIDIA and AI trends more broadly, as well as the importance of improving and modernising legal and compliance structures for public biological data sets.

You can listen to episode 117 of the pharmaphorum podcast in the player below, download the episode to your computer, or find it - and subscribe to the rest of the series - in iTunes, Spotify, acast, Stitcher ,and Podbean.