Amy X. Lu

Tokenized and Continuous Embedding Compressions of Protein Sequence and Structure

The paper compresses protein sequence and structure representations, showing that compressed embeddings retain structural accuracy and enable downstream applications.

Genome modeling and design across all domains of life with Evo 2

A 40B-parameter DNA model trained on 9.3 trillion tokens scores coding and noncoding variants zero-shot, exposes interpretable sequence features, and writes chromosome-scale sequences.

Controllable All-Atom Protein Generation with Latent Diffusion

PLAID generates controllable all-atom protein structures from sequences using a diffusion model trained on sequence data, enabling functional control through Gene Ontology and organism keywords.

Protein Language Model Fitness Is a Matter of Preference

This study investigates how the likelihood of protein sequences affects the zero-shot fitness prediction capabilities of protein language models.

Strategies for effectively modelling promoter-driven gene expression using transfer learning

This paper develops transfer learning strategies to model promoter-driven gene expression in understudied cell types, showing improvements in performance with pretraining and transfer.

Discovering molecular features of intrinsically disordered regions by using evolution for contrastive learning

The paper introduces a method to discover molecular features of intrinsically disordered regions using contrastive learning with evolutionary conservation.

Self-Supervised Contrastive Learning of Protein Representations By Mutual Information Maximization

CPCProt pretrains protein embeddings by telling a protein's own next fragment from fragments of random proteins, reportedly matching benchmarked self-supervised models with 2-10% of their parameters.