research

mission

Life is implemented by the interactions between proteins and other biomolecules. We are in the dawn of a new era of structural biology where machine learning is actively being used in real applications to predict, visualize, and design these biological molecules. Our group develops algorithms that can leverage new sources of data, especially experimental cryo-EM imaging data, to probe challenging new areas of structural biology with impactful applications across basic biology, human health, and bioengineering.

research

Areas of interest include:

• multiview reconstruction in cryo-electron microscopy (cryo-EM),
• 3D representation learning for proteins from sequence, structure, and imaging data,
• generative modeling of protein molecular dynamics, and
• cryo-ET image processing algorithms for in situ cellular visualization.

Computational problems in protein biology pose unique challenges requiring the development of novel methods across many domains of AI research including geometry, vision, and language. We believe that methods motivated by structural biology applications can inspire algorithms of general interest in machine learning, and we seek to make cross-cutting connections across areas. Our research program also emphasizes collaboration with experimentalists to interpret data from the latest experimental tools in structural biology for impactful applications in scientific discovery and engineering.