We have developed WAND (Westlake AI for Novel Drug), an AI-driven drug discovery platform, by integrating physics-inspired AI and cutting-edge theories and technologies across multiple disciplines in drug R&D. WAND spans key stages of drug discovery, from target identification and hit discovery to lead optimization. It makes drug discovery more precise, efficient, and cost-effective.

Our strengths in model, data and computing power ensure the reliability of WAND.

MODEL

WAND is based on a physics-inspired concept. By embedding physical laws as prior knowledge and hard constraints into the AI models, we reduce overfitting risks in data-scarce scenarios and enhance AI prediction reliability.

DATA

The platform integrates high-quality data resources, including our high-accuracy in-house experimental datasets, 10-billion-scale virtual and physical compound datasets, multi-omics datasets from Westlake University. This large-scale and specialized data system provides a solid foundation of our AI models.

COMPUTING POWER

By combining the computing resources of Westlake University and Westlake Laboratory, and our high‑performance cluster, we have established an industry‑leading hybrid architecture for WAND. It enables powerful parallel computing and supports customized workflows for drug discovery.

Grounded in a deep understanding of life sciences, our extensive interdisciplinary research experience and accumulated expertise in CADD form the powerful engine behind WAND. The platform, now applied across our pipeline, enables higher efficiency than traditional approaches.