A grid of evenly spaced, small white dots arranged in a circular pattern on a light gray background. The dots gradually fade out toward the edges.

Heterogeneity in Cryo-EM

We develop data analysis techniques that decrypt the results of cryo-electron microscopy.

Background

In the last decade, cryo-EM has become a premier tool for achieving high resolution structures of biomolecules, but the ability to access the thermodynamics of biomolecules by measuring conformational landscapes within cryo-EM images remains a rich area for developing new methods.

The Structural and Molecular Biophysics team is uniquely positioned to tackle this problem, combining the Flatiron Institute’s computational resources needed to run large-scale molecular dynamics simulations with the technical expertise needed for model-based Bayesian approaches to this central inverse problem in cryo-EM.

Our developments in this area include the ensemble reweighting method, in which the ensemble density present in cryo-EM samples is inferred directly in conformational space by reweighting an ensemble generated from simulations like molecular dynamics.

Data Challenge

The Inaugural Flatiron Institute Cryo-EM Conformational Heterogeneity Challenge

In June 2023, we provided two single-particle datasets with continuous heterogeneity, providing the opportunity to compare results of various heterogeneity analysis methods on identical datasets. Please use your method of choice to analyze these data and provide results in the form of 80 volumes (at 2.146 Å/voxel aligned to the reference volume) with the relative population of each, and their coordinates along the dominant degree of freedom (finely sampling the direction of greatest conformational change). We are providing angles with the datasets, but note that they are only provided as a starting point for the analysis. We expect better results can be obtained with further refinement.

The initially released datasets had 33 742 particles each, they were designed to be small and robust for rapid methods development. Upon feedback from the community, there was also considerable interest in studying a dataset with many more particles. Hence, we have produced a dataset with the same characteristics as the second dataset but many more images (674840).

This expanded dataset can be downloaded using Globus (.tar, 541 GB; Globus login required).

The 2023 challenge has now closed and the preprint is now available at bioRxiv. Additionally, the datasets remain available here:

Download dataset 1 (.tar, 25.2 GB)
Download dataset 2 (.tar, 37.8 GB)