If you work anywhere near virology, structural biology, or computational genomics—and especially if you’ve ever wished for a true map of the SARS-CoV-2 spike protein’s mutational landscape—this episode and the new FAIR² Data Article from the Stay Ahead Project is for you.

We’re diving into the Stay Ahead Project’s latest data release: a thoroughly curated, structure-informed resource focused on the spike protein’s receptor binding domain (RBD). Created by a team led by Erik Schultes (LACDR/GoFair Foundation) with Max Van de Boom and Thomas Hankemeyer, this dataset systematically catalogs every possible single-point mutation in the RBD—over 3,700 in total—plus real-world Omicron variants.

What sets this resource apart? It combines state-of-the-art protein structure prediction (using both AlphaFold2 and ESMFold), deep mutational scanning data for ACE2 binding and surface expression, and biophysical sequence features. We discuss the technical details, the challenges of integrating computational and experimental data, and how this dataset can inform predictive modeling of variant behavior.

We also talk openly about the limitations—like the focus on the RBD and the challenges of modeling higher-order mutational effects—and the ways this resource, accessible via the FAIR² Data Portal, can support the research community moving forward.

If you’re looking for new tools for variant surveillance, functional annotation, or just want a deeper understanding of spike protein evolution, this episode is for you.

van den Boom M, Schultes E and Hankemeier T (2025) Structure-based prediction of SARS-CoV-2 variant properties using machine learning on mutational neighborhoods. Front. Bioinform. 5:1634111. doi: 10.3389/fbinf.2025.1634111

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit senscience.substack.com