Frank J. Poelwijk

Illuminating protein space with a programmable generative model

This paper introduces Chroma, a generative model for proteins that can sample novel structures and sequences, conditioned on desired properties and functions.

Learning the pattern of epistasis linking genotype and phenotype in a protein

The paper studies epistasis in a fluorescent protein by analyzing mutations linking two phenotypically distinct variants, revealing high-order interactions and sparse patterns.

3D protein structure from genetic epistasis experiments

The paper uses genetic epistasis experiments to predict 3D protein structures with 86% accuracy for residue contacts and generates models within 2.4 Å of known structures for small proteins.