MIT

A constrained deep flow-matching framework for distributional translation of omics signatures across biological domains, such as mouse-to-human transcriptomics, without paired samples.

Lightweight multimodal foundation model integrating spatial transcriptomics and H&E histopathology with pathway activity scores for biologically grounded spatial niche discovery at single-cell resolution.

A self-supervised, cell-centric pretraining strategy that distills morphology and microenvironment views of each cell into a unified embedding for virtual spatial omics from microscopy.

Energy-based model of protein conformational space derived from diffusion-model likelihoods, usable as a universal statistical potential for many tasks.

Open model that jointly predicts biomolecular structure and small-molecule binding affinity, approaching FEP+ accuracy in seconds on a single GPU.

Open multimodal clinical foundation model that jointly reasons over medical images, ECG time-series, and text reports, trained with domain-aware reinforcement learning.

A self-supervised motion foundation model for wearable accelerometry, trained with relative contrastive learning on 1B segments from 87,376 participants.

Open-source deep learning model for biomolecular structure prediction achieving AlphaFold3-level accuracy, trained entirely on publicly available data.

Interactive foundation model for biomedical image segmentation, prompted with scribbles, clicks, and bounding boxes to segment unseen structures.

A foundational keypoint model for robust, flexible brain MRI registration, pretrained on over 100,000 3D volumes and supporting rigid, affine, and deformable alignment.

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

AlphaFold fine-tuned with flow matching to generate conformational ensembles, capturing protein dynamics and flexibility beyond single static structures.

A rigorous benchmarking study of scBERT and scGPT for cell type annotation, comparing foundation models against logistic regression baselines.

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