Research | Dr. Oleksandra Kukharenko
Improving and extending regression techniques
We use data-driven techniques of statistical/machine learning in order to address some of the problems that theoretical/computational chemistry is facing in the context of polymer studies. Examples are: enhance sampling and coarse graining, back-mapping, state’s identification and characterization.
Feature importance analysis
We focus on inferring of importance of the input features for the determination of physically/chemically meaningful descriptors in the data. It is of great importance in solving the problem of ill conditioning of the regression models, as well as for improving transparency of nonparametric prediction models.
SFB 1551
We have two collaborative projects (R02 and R03), where we focus on development and refining of coarse-grained models that can accurately and efficiently capture phase dynamics of multi-component systems of large biopolymers. We characterize and model effects of ubiquitylation and sumolation on solubility, phase separation, and aggregation behavior of target proteins. We examine the behaviour and dynamics of branched and linear polymer chains. We complement coarse graining with back-mapping to validate underlining models.