Vanderbilt Data Science is excited to feature the impressive work of postdoctoral fellow Zhaoqian (Joshua) Su, who is leading multiple projects that apply generative AI to revolutionize protein science. These projects, all conducted under the guidance of our faculty, showcase how artificial intelligence can address complex biological challenges.
SuperMetal: Predicting Metal Ion Locations in Proteins
Researcher: Xiaohan Kuang (Vanderbilt DSI Master’s Student)
Many proteins require metal ions to function. SuperMetal leverages generative AI to predict the exact locations of metal ions in protein structures, providing insights crucial for metalloprotein engineering. This framework has achieved state-of-the-art performance, with a manuscript already submitted for publication.
SuperWater: Mapping Water Molecules in Protein Structures
Researcher: Xiaohan Kuang (Vanderbilt DSI Master’s Student)
In biological systems, water molecules are essential to maintaining protein structure and function. SuperWater uses generative AI to predict water positions on protein surfaces, offering new avenues for drug design and structural biology. This project, with its results currently being prepared for publication, holds promise for applications in molecular docking and water-mediated interactions.
De Novo Peptide Design Using Generative AI
Researcher: Hexuan Fan (Vanderbilt DSI Master’s Student)
The De Novo Peptide Design project explores how generative AI can predict peptide-protein interactions, accelerating peptide drug development. With applications in cancer, metabolic disorders, and infectious diseases, this work could dramatically reduce the time and cost associated with traditional experimental approaches. The team is currently in the process of model training, with a grant proposal already submitted.
Epitope Prediction with Physics-Informed Neural Networks
Researcher: Yuhao Zhang (Vanderbilt DSI Master’s Student)
Combining deep learning and protein-protein interaction data, the Epitope Prediction project seeks to aid in antibody design and therapeutic discovery. By incorporating physics-based models, this approach enhances the accuracy of predictions, with validation and manuscript preparation ongoing.
Discovering Amyloid-Like Peptides Through AI and Molecular Simulations
Researcher: Xiaohan Kuang (Vanderbilt DSI Master’s Student)
This innovative study integrates molecular dynamics simulations with AI to explore the relationship between peptide sequences and their propensity to form amyloid fibrils. The project has broad applications in the treatment of neurodegenerative diseases such as Alzheimer’s and Parkinson’s, with results to be published soon.
Vanderbilt Data Science continues to support these transformative research efforts, driving the use of AI in tackling some of the most challenging questions in modern biology.