Projects

Computational Instruments

Three tools, three scales, one framework — the AI Virtual Cell.

01
Cellular Scale

TME-VI

Tumor Microenvironment Variational Inference. An uncertainty-aware decoder virtual instrument that combines scVI cellular representations with a Graph Attention Network and Monte Carlo Dropout to predict immune escape and anti-PD-1 immunotherapy response in melanoma patients.

scVI GAT MC Dropout GSE120575 16,290 cells 32 patients
0.885 Cell AUROC
0.97 Filtered AUROC
0.78 Patient AUROC
HF Space ↗
02
Molecular Scale

MutaGraph

GNN-based protein mutation effect prediction platform. Predicts how single-point mutations affect protein stability (ΔΔG) and antibody-antigen binding affinity using graph attention over protein structure graphs.

PyTorch Geometric GAT ΔΔG Protein Structure Binding Affinity
HF Space ↗
03
Knowledge Scale

CellScout

AI research assistant for the AIVC literature. Transforms academic papers into a queryable knowledge graph, enabling researchers to ask natural-language questions and receive cross-paper synthesis powered by a multi-agent RAG system.

Multi-agent RAG Knowledge Graph Neo4j Claude API
HF Space ↗