AIDDPM Lab — AI for Drug Discovery & Precision Medicine

Research Directions

What We Work On

AIDDPM Lab sits at the intersection of artificial intelligence, computational chemistry, and biomedical science. Our four core research pillars drive discoveries from molecular design to clinical translation.

💊

AI-Driven Drug Design & Discovery

Deep learning models for molecular property prediction, lead optimization, and closed-loop automated drug design platforms. Key tools: MolMap, Leadmaster, and contrastive learning for activity cliff overcoming.

🤖

AI Agent for Drug & Target Discovery

Building autonomous AI agent systems that close the loop between molecular design, synthesis planning, and experimental feedback. Includes multi-step reasoning agents for automated hit-to-lead acceleration and target identification.

🎯

Precision Cancer Immunotherapy

Building predictive models for immunotherapy response across cancer types. Our COMPASS model achieves multi-center clinical validation and drives clinical translation of AI in precision oncology.

🧬

Cancer Multi-Omics AI

Integrating genomics, transcriptomics, and proteomics through structured representation learning to uncover tumor heterogeneity and identify novel therapeutic targets. Key tool: AggMap.

🔭

Foundation Models for Biomedicine

Developing and fine-tuning large-scale pre-trained models for molecular and clinical data. Leveraging multi-modal fusion, contrastive learning, and retrieval-augmented generation to advance drug target identification and therapeutic hypothesis generation.

Lab Updates

News & Highlights

2026.03 PreprintFPPool — a novel knowledge-guided pooling framework bridging molecular fingerprints and graph neural networks — is now on ChemRxiv. Achieves state-of-the-art on 40 benchmarks with built-in interpretability (Cui, Feng, Song, Yu, Lu, Chen YZ, Shen W.).
2026.01 PreprintMedea, an omics AI agent for automated therapeutic discovery, is posted on openRxiv. The agent integrates reasoning over multi-omics data to identify and prioritize disease targets (Sui, Shen et al., Zitnik lab).
2025.07 WelcomeDr. Wanxiang Shen joins Zhejiang University as Assistant Professor under the ZJU Hundred Talents Program (百人计划). He was also selected for the National High-Level Overseas Young Talent Program.
2025.05 PreprintCOMPASS — generalizable AI for predicting immunotherapy response across cancer types and treatments — is under review (Shen, ..., Zitnik et al.).

All Publications →

Open-Source Software

Our Tools

Production-ready AI tools for molecular science, freely available to the community.

COMPASS

Immunotherapy response prediction · Multi-center validated

LAUNCH →

MolMap

2D molecular feature maps for CNN-based drug property prediction

GITHUB →

AggMap

Multi-omics data structurization and deep learning integration

GITHUB →

Leadmaster

Automated closed-loop drug design platform

LEARN MORE →

MusicMol

Transform molecular structures into music — chemistry made audible

LAUNCH →

Institutional Connections

Global Collaborations

🏛 Zhejiang University 🔬 Harvard Medical School 🌐 National University of Singapore 💊 Roche pRED 🎓 Tsinghua University 🤖 ZairaChem 🏛 Shenzhen University

AI-Driven Drug Discovery & Precision Medicine