Harvard University

Transformer framework that models protein-protein interactions at residue resolution, generalizing zero-shot to unseen MHC alleles and sequence-neutral PTMs from one fixed checkpoint.

Channel-adaptive autoregressive generative model that synthesizes in-silico fluorescence microscopy of protein subcellular localization from amino-acid sequence and cellular landmark stains.

An RNA language model trained by self-supervised masked-nucleotide prediction on ~50,000 IRES sequences that predicts secondary-structure features rivaling experimental chemical probing.

Geometric deep learning model that learns universal atomic-scale representations of intermolecular interfaces across proteins, small molecules, ions, lipids, and nucleic acids.

A full-atom SE(3)-equivariant diffusion model that inpaints binding interfaces to design proteins that bind DNA, RNA, and small molecules.

A native 3D vision transformer self-supervised on unlabeled fluorescence microscopy volumes that generalizes to unseen object classes without retraining or voxel annotations.

Predicts single-cell-resolution scRNA-seq coverage and scATAC-seq insertion profiles directly from DNA sequence by adapting the Borzoi predictor with a cell-specific decoder.

A modality-agnostic, multi-task foundation model for human brain imaging that runs five core tasks across uncalibrated CT and MRI without retraining.

Universal foundation model that jointly generates diagnostic text and segments the corresponding targets across ten biomedical imaging modalities.

A vision-language foundation model for precision oncology that pretrains on 50M pathology images and 1B text tokens via unified masked modeling.

A self-supervised vision foundation model for structural brain MRI that produces general-purpose features adaptable to diverse downstream clinical and neuroscience tasks.

Protein language model for per-residue domain annotation, pairing an ESM-2 backbone with a probabilistic decoder to rival HMMER on sequence domain assignment.

A 1D convolutional foundation model for electrocardiogram analysis, trained on over 10 million expert-annotated recordings across 150 diagnostic categories.

A weakly supervised pathology foundation model pretrained on 60,530 whole-slide images across 19 anatomical sites for cancer detection, prognosis, and molecular prediction.

A medical multimodal large language model pretrained on the 25M-image MedTrinity-25M dataset, achieving state-of-the-art accuracy on biomedical visual question answering.

Geometric deep learning model generating context-aware protein representations across 156 cell-type contexts from a multi-organ single-cell atlas.

A multimodal vision-language copilot for human pathology that analyzes histology images and answers diverse pathology queries in natural language.

Genomic language model trained on metagenomic scaffolds that learns protein co-regulation and function by modeling gene context and operon structure.

A self-supervised 3D CT foundation model that extracts general-purpose tumor representations for cancer imaging biomarker discovery across diverse downstream tasks.

A multimodal medical vision-language model that performs few-shot generative visual question answering over medical images and text.

An in-context learning model that segments unseen medical imaging tasks from a few labeled examples, with no retraining or fine-tuning.

A multiplicative LSTM protein language model trained on 24M sequences to produce fixed-length embeddings for protein engineering and function prediction.

Deep convolutional neural network that learns the regulatory code of DNA accessibility from DNase-seq data across 164 cell types, enabling variant effect prediction at cis-regulatory elements.