Title:	Analysis of cryo electron microscopy data with artificial intelligence
DNr:	Berzelius-2025-39
Project Type:	LiU Berzelius
Principal Investigator:	Lukas Grunewald <lukas.grunewald@kemi.uu.se>
Affiliation:	Uppsala universitet
Duration:	2025-01-31 – 2025-08-01
Classification:	10610
Homepage:	https://www.westenhofflab.net/
Keywords:

Abstract

Machine learning models can be used to complement experimental data and proved to be a powerful tool to aid to solve and validate structures and dynamics of proteins. By combining machine learning with experimental data, they can prove previously unexplained changes and behavior in proteins. For my project, I would like to use alphafold for protein structure prediction and furthermore, train my own neural network for analysing conformational variability in cryo-electron microscopy and use this information to reconstruct protein ensembles. cryoDRGN will be used as well to analyse and compare our method to it. We are currently under revision for our paper, but we still need to improve and revise our code. For that we need to rerun certain experiments. Furthermore, we have a few more ideas of how to apply machine learning to experimental data like time-resolved X-ray crystallography.