We’re taking a look at ProtRL: a framework for aligning protein language models to your desired distributions using reinforcement learning. Filippo Stocco is telling us all about how it works, why reinforcement learning is important for protein engineering, and how these proteins perform when tested in our lab at Adaptyv.

In this designer spotlight, we take a look at Michael Hla’s recent Pro-1. It is a protein reasoning and optimization model capable to explain why it proposes a mutation. He tested FGF-1 designs in our lab at Adaptyv and managed to show significant melting temperature improvements, while still maintaining binding to their targets. One design even reached a melting temperature comparable to the most optimized FGF-1 variants in the literature!

We wrote a community paper about our Protein Design Competition, teaming up with your favourite protein designers from both rounds. We close the paper by creating BenchBB, the Bench-tested Binder Benchmark — a curated set of 7 protein targets designed to capture diverse binder design challenges by remaining accessible enough for wide scale lab validation

In this case study, we highlight how Microsoft Research used our automated lab to validate proteins generated by EvoDiff, their novel sequence-first protein design model, in just a few weeks.

We analyze the results of our protein design competition where 130 designers created binders for EGFR. With a 5x improvement in success rates and some designs outperforming clinical antibodies, we explore what worked, what didn't, and what this means for the future of protein engineering

What a close race this one was! In this blog post, we look at how submissions evolved throughout our latest EGFR binder design competition. We highlight the most widespread model and design choices, how some people exploited loopholes in our scoring system, and argue if de novo design is reliable or not. We also have plenty of animations for you to click through!

Imagine if you could let your AI agent design novel proteins, autonomously test them in our wet lab and then improve itself based on the results.

The results of the first round of the EGFR binder design competition are here! In this blog post, we analyze the most common model and design choices, take a closer look at the strategies that yielded successful binders, and provide recommendations for your future design campaigns.

For this technical blog post, we surveyed the state of the art of using ML for protein optimization. We focused on adaptive, cost constrained, and multi-objective methods and have summarized the best approaches for you.

Using our automated Affinity Characterization workflow we validated RFDiffusion designed binders in less than 24h