Mila

Multimodal diffusion model that co-designs protein sequence and 3D structure around arbitrary biomolecules, demonstrated by designing novel heme enzymes catalyzing carbene-transfer reactions.

A generative foundation model for biomolecular dynamics that produces atom-level MD-style trajectories for protein monomers and protein-ligand complexes.

Property-driven protein inverse folding: a ProteinMPNN checkpoint aligned via multi-objective preference optimization to improve developability while preserving structural fidelity.

Functional diffusion model that predicts single-cell perturbation responses by generating over distributions embedded in a Hilbert space, capturing population-level response variability.

A family of autoregressive genomic foundation models that reconcile k-mer tokenization with single-nucleotide resolution at contexts up to 98k bp.

A SimCLR self-supervised foundation model for 3D brain MRI, pretrained on 18,759 patients across 11 neurological-disease datasets for diverse diagnostic tasks.

A joint sequence-structure representation learning framework combining ESM-2 protein language model embeddings with GearNet geometric graph neural networks.

A geometric relational graph neural network that learns protein structure representations via geometry-aware message passing and self-supervised pretraining.