Approximating Claude's protein mutation response with an alanine scan

sokrypton · x · 2026-11-18

sokrypton replies to @proteinrosh: getting the logits would require multiple runs per site to build a distribution, including responses to every mutation — too expensive (he's out of tokens for the week). He suggests approximating with an alanine scan, i.e. using his earlier Jacobian method to estimate Claude's response to mutation effects and predict whether it can understand long-range protein contacts.

Related event: Researcher tests whether Claude Fable can predict protein contact maps from sequence alone(12 posts)→

Original post →

More from Models

Models channel →