Seminar: Erik Thiede

The Simons Center for Computational Physical Chemistry at NYU regularly hosts visiting scholars to discuss their work.  Join us on January 13th for a presentation by Erik Thiede of the Cornell University:

Recovering the Conformational Ensembles from Cryo-EM using Gradient Flows

Abstract:
One of the biggest challenges in biochemistry is understanding the ensemble of conformations that proteins adopt. Cryogenic-sample Electron Microscopy (cryo-EM) gives us a unique opportunity to recover this ensemble by letting us image individual conformations, albeit in extremely noisy images. To recover the conformational ensemble in the presence of extreme noise – cryo-EM images typically have signal-to-noise ratios of 0.1 or lower – we introduce a family of gradient flows that can recover the latent probability density over conformations. Our formalism gives a principled way of recovering conformational ensembles from noisy single-particle data, along with theoretical guarantees. Moreover, a specific hyperparameter choice and discretization recovers 3D classification: one of the most commonly used algorithms for analyzing heterogeneity in cryo-EM. Applying our analysis shows that the common practices of interpreting the resulting structures as “classes” or counting how many images match a structure are mathematically dubious. But by reconsidering the role that the structures play, we show how 3D classification can potentially recover conformational probability densities.

All Simons Center seminars are held in Waverly 540.  Refreshments will be served at 10:45, and the seminar begins promptly at 11:00 AM ET.      

Or join via Zoom:  https://nyu.zoom.us/j/99318701420?pwd=eGVvSzlKWFRlV0ZldnJJbjhYVUtEQT09