Nicholas Bhattacharya

This paper introduces a self-supervised learning approach to detect and classify biosynthetic gene clusters in microbial genomes using masked language models trained on protein domains.

This paper introduces a differentiable Smith-Waterman algorithm to jointly learn multiple sequence alignments and downstream tasks, showing improvements in contact prediction and structure prediction.

The paper shows that a single attention layer can predict protein contacts as well as Potts models and pretrained Transformers, suggesting attention models can capture meaningful protein features from unlabeled data.

The paper uses transporter gene frequencies to predict metabolite characteristics and identifies siderophore-like BGCs in gut microbiomes.

Across five protein tasks with evolution-aware splits, self-supervised pretraining helped nearly every model on every task, yet alignment-based features still beat learned embeddings on secondary structure and contact prediction.