Artificial Intelligence (AI) in Protein Structure Prediction and Design

Research Group Lead PI: Hsuan-Yu Chen (陳璿宇)

Mission & Research Objectives

Our mission is to improve protein structure prediction and design models, specifically focusing on cancer-related mutant proteins.

Core Objectives:

Research Plan & Milestones

Our three-year research strategy is divided into progressive phases aimed at clinical integration and model optimization.

Phase 1: Initial Modeling & Validation

Focus on critical mutant proteins including KRAS G12X and EGFR C797S. Use cryo-EM data to refine the AlphaFold backbone and validate biological function in vitro and in vivo.

Phase 2: De Novo Antibody Development

Utilize the Rosetta model to predict antibodies targeting validated mutant structures. Conduct binding affinity tests and integrate this feedback to optimize the design model.

Phase 3: Model Tuning & Scaling

Apply learned optimization "tips" to extend predictions to other important mutant proteins. Establish a robust pipeline that requires fewer physical experiments for new targets.

Core Innovation

"Our core idea is to integrate information from current established AI models with real cryo-EM and biological experiments to develop AI models with high accuracy and to provide solutions to clinical unmet needs."

By treating experimental data as a new component in the learning model—similar to the impact of Multiple Sequence Alignment (MSA) in previous breakthroughs—we aim to transform how AI models are tuned for precision medicine.

Research Team

Role Name Affiliation
Research Group PI陳璿宇 (Hsuan-Yu Chen)Institute of Statistical Science (ISS)
Co-PI杜憶萍 (I-Ping Tu)Institute of Statistical Science (ISS)
Co-PI袁新盛 (Shin-Sheng Yuan)Institute of Statistical Science (ISS)
Co-PI章為皓 (Wei-Hau Chang)Institute of Biological Chemistry (IoC)
Co-PI陳玉如 (Yu-Ju Chen)Institute of Biological Chemistry (IoC)
Co-PI張雅媗 (Ya-Hsuan Chang)National Health Research Institutes (NHRI)
Co-PI黃國彥 (Kuo-Yen Huang)National Taiwan University (NTU)

References