Purdue University

Protein language model that supervises embedding geometry with inter-residue contacts, so representation distance tracks physical distance.

Few-step generative model for de novo protein backbone design, extending MeanFlow to the SE(3) Lie-group geometry of residue frames.

Protein conformational ensemble generator and coarse-grained force field in one normalizing flow. Samples faster than diffusion-based baselines.

Molecular representation learning from surface point clouds, 3D graphs, and fragment tokens. Cuts ESOL RMSE to 0.740 under scaffold splitting.

Structure-based drug design language model fusing protein structural and evolutionary encoders with SAFE fragment tokens for hit-to-lead generation.

Cryo-EM ligand modeling pipeline that detects bound ligand densities in a map, then reconstructs their atomic structures with a diffusion model.

Multimodal model that designs small molecules from transcriptomic and cell-imaging perturbation phenotypes with a rectified flow transformer.

Protein language model that emulates molecular dynamics, generating equilibrium conformational ensembles and multi-timescale dynamic trajectories.

Sequence-based multitask model predicting covalently ligandable cysteines and reversible ligand-binding residues across the human proteome.

Protein structure retrieval model aligning 3D structures with functional text via contrastive learning, for zero-shot search of PDB and cryo-EM maps.

End-to-end transformer reading fluorescence microscopy video to return single-molecule trajectories with Hurst exponents and diffusion coefficients.

Medical image segmentation model that replaces MedSAM's manual box prompts with a diffusion prompt encoder and labels each mask by class.

Cryo-EM reconstruction with neural radiance fields in Euclidean 3D space, separating conformational motion from compositional assembly states.

Hi-C foundation model pretrained on 118 million contact submatrices, fine-tuned for loop detection, resolution enhancement and epigenomic prediction.

Protein conformational ensemble generation guided by experimental observables, steering a pretrained diffusion sampler toward Boltzmann statistics.