Learned models for molecular reconstruction in cryogenic electron microscopy

Cryogenic electron microscopy (cryo-EM) is a powerful imaging technique for reconstructing 3D models of biological macromolecules using transmission electron microscopy.

Although this method is now able to reach near-atomic resolution for certain molecules many important challenges remain. Foremost is the issue of noise.

Due to the low electron dose required to reduce specimen damage, cryo-EM images are extremely noisy, a fact that is exacerbated for small molecules that have a low signal power to begin with.

As a result, only molecules above a certain size can be consistently imaged using cryo-EM and very large datasets are required to obtain a reasonable accuracy.This project instead proposes to mitigate the high noise level using priors on the 3D molecular structures.

These will be learned from data using previously obtained molecular structures as well as structures synthesized by tools like AlphaFold 2.

More specifically, by training deep neural networks (DNNs) to estimate various quantities along the cryo-EM reconstruction pipeline, we can encode the natural structure of the data and reduce the effect of the noise.

The resulting method will allow for more accurate reconstructions at high noise levels, reducing the expense of data collection and enabling reconstruction of small molecules.

Finally, the method will also enable accurate and robust reconstruction of structural variability from moderate-sized datasets, something which is not possible using current methods.