Protein Structure
Learn how modern generative models represent and sample protein geometry for structure prediction, design, and biological discovery.
ENAR 2026 Spring Meeting
Generative Models for Protein Structures and Biomedical Data
SC3 short course covering core generative modeling methods and their use in protein structures, drug design, single-cell trajectories, and biomedical data with practical demos.
Interactive PDB structure · 1HV4 · HIV-1 Capsid
Course highlights
Featured Modules
Three application areas anchor the course: structural biology, molecular design, and temporal biomedical data.
Drug Design
Explore generative strategies for molecule discovery, candidate optimization, and search over large chemical design spaces.
Time Series
Cover diffusion, sequence modeling, and trajectory analysis for longitudinal signals, single-cell transitions, and other biomedical time-dependent data.
Course structure
Syllabus
Expanded outline for quick access during and after the course.
Section 1: Foundation Models
- Classic generative models
- Autoregressive models
- VAE and GAN
- Normalizing flow
- Diffusion and flow matching
- Discrete diffusion model
- Code demo: vibe coding
Section 2: Biological Applications
- Proteins
- Drug design
- Single-cell trajectory
- Code demo
Section 3: Biomedical Data
- Biomedical data challenges
- Statistical tasks: imbalanced learning
Resources
Course Materials
Slides are ready to attach now. Recording and code links can be added as they become available.
Section 1 Materials
Section 2 Materials
Section 3 Materials
People
Instructors
Anru Zhang
Duke University
Primary faculty jointly appointed by Biostatistics and Bioinformatics, and Computer Science at Duke University.
Read more
Anru received his PhD from the University of Pennsylvania in 2015. He is the recipient of the IMS Tweedie Award, COPSS Emerging Leader Award, and ASA Gottfried E. Noether Junior Award. His work is currently supported by NIH R01 grants and an NSF CAREER Award.
Alex Tong
Aithyra, Vienna
Principal Investigator at Aithyra, formerly (briefly) assistant professor at Duke University.
Read more
Previously, Alex was briefly an assistant professor at Duke University. He completed a postdoc with Yoshua Bengio at Mila and earned his PhD from Yale University in 2021. His research spans generative modeling, deep learning, optimal transport, and causal discovery for cell development, molecular biology, and protein design. He co-founded Dreamfold.
Fred Peng
Duke University
PhD student at Duke University focused on generative modeling and protein design.
Read more
Fred works on practical and theoretical advances in generative models for protein structures and biomedical data applications.