Sai T. Reddy

GenLoop uses dtSFM and AlphaFold 3 to iteratively design small-molecule drugs, achieving high structural confidence and resolving paralog selectivity across multiple targets.

This paper introduces PGLD, an AI method that designs cost-effective DNA libraries for high-throughput experiments by optimizing synthesis-aware stochastic libraries.

This paper introduces dtSFM, a foundation model for drug-target interactions that predicts off-targets, identifies drug candidates, and generates novel molecules using thermodynamics-based sequence modeling.

The paper introduces Specificity Foundation Models (SFM) that predict molecular binding across six domains using a physics-inspired architecture, achieving high retrieval accuracy without structural data.

The paper uses deep mutational learning to analyze serum antibody escape to SARS-CoV-2 variants, finding individualized escape profiles and conserved binding signatures.

The paper establishes a neural architecture for molecular recognition inspired by physics principles, achieving high data efficiency in antibody-antigen retrieval.

This paper introduces CALM, a model that predicts antibody-antigen binding from sequences using contrastive learning, achieving 7% top-1 retrieval on a test set.

This paper develops DROP-CARs, drug-regulated CAR-T cells that control cell-cell interactions using small-molecule switches, showing in vitro and in vivo functionality.

The paper shows that transformer attention and contrastive learning can be mathematically derived from immunological principles, highlighting computational convergence between adaptive immunity and AI.

This study designs antibody Fc-variants with programmable functional profiles using deep learning and experimental validation.

This paper introduces a yeast mating-based platform for generating ultra-large antibody libraries and demonstrates its ability to discover high-affinity binders.

AntibodyForests quantifies inter- and intra-antibody repertoire evolution using protein language models and structure analysis.

The paper develops a high-throughput method to profile SARS-CoV-2 RBD variants' affinity against antibodies, generating binding energy landscapes for variant analysis.

This study uses synthetic coevolution to explore mutational trajectories between neutralizing antibodies and SARS-CoV-2 variants, revealing antagonistic and compensatory mutations.

The paper characterizes clonal dynamics of antibody-secreting cells in bone marrow following influenza infection, finding that IgG ASCs preferentially target nucleoprotein regardless of infection type.

The paper presents a synthetic immune cell platform integrated with machine learning to predict TCR specificity to peptide-HLA antigens across the human proteome.

This paper presents a nanovial-based platform for high-throughput functional screening of CAR T cell libraries, identifying CAR constructs with distinct secretion profiles over time.

PLMFit benchmarks transfer learning methods on protein language models for protein engineering tasks, identifying optimal strategies for knowledge transfer.

This study shows that protein language model pseudolikelihoods correlate with B cell selection features like SHM and isotype usage, but not binding affinity, in vivo.

This paper introduces a high-throughput method for screening antibody neutralization using droplet arrays, demonstrating scalable functional antibody data collection against viral diseases.

This study uses antibody repertoire data and machine learning to predict and engineer antibody affinities, achieving accurate predictions and successful in silico design of seven out of eight synthetic variants.

This paper uses conditional optimal transport to model CAR T cell responses at the single-cell level and predict gene expression for unseen CAR variants.

This paper introduces TouCAN, a contrastive learning method for clustering TCR sequences and predicting antigen specificity, showing comparable performance to existing methods.

This paper introduces nucleotide augmentation (NTA) to improve machine learning models for protein engineering by generating biologically meaningful sequence data, achieving performance gains even with limited training data.

The paper introduces immuneREF, a method to quantify immune repertoire similarity across health and disease, finding that healthy and diseased repertoires are more similar than previously thought.

The paper identifies and validates two novel human genomic sites for safe and predictable gene expression in therapeutic applications.

This paper develops deep mutational learning to predict the impact of SARS-CoV-2 spike protein mutations on ACE2 binding and antibody escape, identifying potential variants and supporting vaccine and therapeutic development.

speedingCARs uses domain shuffling and single-cell sequencing to engineer CAR T cells, identifying effective variants with diverse transcriptional signatures and T cell phenotypes.

This study examines how different computational methods affect the analysis of B cell evolution in murine antibody repertoire sequencing data post-viral infection.