The motion of a protein is very much connected to the function of a protein. Being able to predict the motion of a protein is therefore an important step in drug design.
Description of Work
The thesis builds on the work from Ines Putz, applying the break through of distogram prediction from the contact prediction field to protein motion. Elastic Network Models are useful in determining the coarse-grained motion of proteins. During structural transitions, certain residue pairs that were spatially close become separated, so called breaking contacts. Incorporating this information improves the prediction of protein motion. In this thesis, we want to apply state of the art machine learning methods to breaking contact prediction and further evaluate distance prediction instead of binary contacts. Here, distograms can provide a way to vary the stiffness of the springs in the elastic network model to allow for more flexibility.