Kadina E. Johnston

FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications

FLIP2 adds seven protein fitness datasets and 16 splits for enzymes, light-sensitive proteins, and interactions; simple ridge and zero-shot baselines often rival fine-tuned protein language models.

LevSeq: Rapid Generation of Sequence-Function Data for Directed Evolution and Machine Learning

LevSeq rapidly generates sequence-function data for protein engineering using nanopore sequencing and dual barcoding, enabling directed evolution and machine learning applications.

Evaluation of Machine Learning-Assisted Directed Evolution Across Diverse Combinatorial Landscapes

This study evaluates machine learning-assisted directed evolution strategies across diverse protein fitness landscapes, finding that focused training with zero-shot predictors outperforms random sampling in challenging landscapes.

A combinatorially complete epistatic fitness landscape in an enzyme active site

The paper constructs a large-scale sequence-fitness dataset for enzyme engineering, revealing significant epistasis and highlighting challenges in predicting optimal variants.