Alvis: Machine learning models have been proven to be powerful tools in aiding the solution and validation of structures and dynamics of proteins and thus enlighten the understanding of molecular functions. Combining these models with experimental results can unravel previously unexplained phenomena in proteins. For this project, we would like to take advantage of alphafold for structure prediction and train a new neural network for analyzing cryo electron microscopy data to identify different protein species on the grid enabling more sufficient data analysis. Furthermore, we need to refine our previous model about protein ensemble generation upon cryo-EM images, which paper is currently under revision. Additionally, there are also some ideas of using machine learning for investigating time-resolved X-ray crystallography. Dardel: After successfully conducting Molecular Dynamics Simulations of different states along the dark reversion of phytochromes determined by cryoEM we want to further explore the phototransition/dark reversion of phytochromes and validate experimental results from cryo electron microscopy and time resolved spectroscopy with the help of enhanced sampling techniques. This would elevate the study tremendously, as it combines computational methods with experiments to shed light on the chemical mechanism.