Ayan Sengupta

A germline shortcut in protein language model retrieval of adaptive immune receptors

This paper benchmarks protein language models against sequence alignment methods for immune receptor retrieval, finding that alignments outperform PLMs at CDR3 level but PLMs match alignments at full variable domain by leveraging germline information.

Accurate and scalable multi-disease classification from adaptive immune repertoires

This paper introduces FACTS, a scalable machine learning framework for multi-disease classification using TCR repertoires, achieving high accuracy and preserving biological signals.

Robust diagnosis of infectious disease, autoimmunity and cancer from the paratope networks of adaptive immune receptors

This paper develops a novel representation of adaptive immune receptors to improve disease classification from blood samples, achieving higher accuracy for infectious disease, autoimmunity, and cancer.