The Use of Inductive Methods for Determination of the Binding Affinity of Interacting Biological Molecules
Iryna Lurie, Volodymyr I. Lytvynenko, Volodymyr Osypcnko, Mariia A. Voronenko · 2018 IEEE 13th International Scientific and Technical Conference on Computer Sciences and Information Technologies (CSIT) · 2018
Binding of peptides plays an important role in the molecular biology. The binding process of peptides can activate the cytotoxic T-cells in the immune system. One of the most difficult aspects of peptide binding process is the prediction of the affinity of protein binding to the peptide. In this paper, we propose a new information technology for the quantitative determination of the affinity of peptide-MHC binding. In the paper a step-by-step method of applying descriptor clustering algorithms has been proposed. At the first stage we use the DBSCAN density algorithm, and the descriptors representing the ghost component with respect to the dependent variable are deleted. At the second stage, a robust set of prognostic features is identified using k-means algorithm. At the final stage, polynomial models are constructed with the use of iterative multilayer algorithm MIA GMDH. The obtained models have high accuracy, robustness, and adequacy.