Member Listing – Hamed Khakzad

Building dynamic-aware deep learning models to predict macromolecular interactions and functions, and to design novel therapeutics

My team (Delta) at Inria aims to develop dynamic-aware data-driven approaches to predict macromolecular interactions and functions, and to design novel functional proteins. While deep learning has advanced static structure prediction, the flexibility of macromolecules remains largely overlooked. We address this gap by building large-scale dynamic databases and creating deep learning models that capture their conformational heterogeneity. Our work focuses on understanding flexible interactions, such as folding-upon-binding, antibody-antigen, and enzyme-substrate complexes, and applying these models to therapeutic design, including protein binders.

Dynamic-aware systems; Deep learning; Protein design

2015