Timezone: »

Physics aware inference for the cryo-EM inverse problem: anisotropic network model heterogeneity, global 3D pose and microscope defocus
Geoffrey Woollard · Shayan Shekarforoush · Frank Wood · Marcus Brubaker · Khanh Dao Duc

We propose a parametric forward model for single particle cryo-electron microscopy (cryo-EM), and employ stochastic variational inference to infer posterior distributions of the physically interpretable latent variables. Our novel cryo-EM forward model accounts for the biomolecular configuration (via spatial coordinates of pseudo-atoms, in contrast with traditional voxelized representations) the global 3D pose, the effect of the microscope (contrast transfer function's defocus parameter), and noise. To capture heterogeneity, we use the anisotropic network model (ANM), a Gaussian in the space of atomic coordinates. We perform experiments on synthetic data and show that the posterior of the scalar component along the lowest ANM mode and the angle of 2D in-plane pose can be jointly inferred with deep neural networks. We also demonstrate Fourier frequency marching in the simulation and likelihood during training, without retraining the neural networks that characterize the variational posterior.

Author Information

Geoffrey Woollard (University of British Columbia)

Geoff earned his BSc (Biophysics, 2011) and MSc (Genome Science and Technology, 2014) at the University of British Columbia. He worked as Associate Scientists for Cyclica and Structura Biotechnology, where he wrote software for structural biology applications, and did wet bench biochemistry and microscopy for single particle electron cryomicroscopy (cryo-EM) studies of a calcium ion channel at the University of Toronto. He is currently doing his PhD at the University of British Columbia, where he works on modelling and inference for cryo-EM.

Shayan Shekarforoush (University of Toronto)
Frank Wood (University of British Columbia)
Marcus Brubaker (York University)
Khanh Dao Duc (University of British Columbia)
Khanh Dao Duc

Khanh Dao Duc received the B.S. degree in mathematics from the Ecole Normale Supérieure de Lyon, Lyon, France, in 2006, the M.S. degree in mathematics from Université Paris 6, Paris, France, in 2009, and the Ph.D. degree in applied mathematics from the Ecole Normale Supérieure, Paris, France, in 2013. From 2014 to 2015, and 2017 to 2019 he was a postdoctoral fellow at the University of California, Berkeley, CA, USA. From 2015 to 2017, he was a Simons postoctoral fellow with the University of Pennsylvania, Philadelphia, USA. He is currently an Assistant Professor of mathematics at the University of British Columbia, Vancouver, BC, Canada, where his research group focuses on developing mathematical and computational methods for various problems in structural, molecular and cell biology, and neuroscience.

More from the Same Authors