ИСТИНА

The key role of molecular modeling and docking programs for the initial stage of the new remedies development is demonstrated. Docking programs position a ligand in the protein and estimate the protein-ligand binding energy. Accurate calculation of the protein-inhibitor binding free energy is the main problem which should be solved to increase effectiveness of computer aided drug design. Performance of docking programs is based on the effective solution of the global optimization problem on the energy surface with large dimensionality. Physical and mathematical problems of development of the new generation of docking programs are briefly discussed. Nowadays due to the technological progress it is possible to yield a lot of unique information about each patient. It is impossible to determine relationships between hundreds or thousands of different patient’s variables without using modern mathematical methods and computers. Probabilistic models based on Bayesian networks are flexible and versatile, they are capable of self-learning and integration of heterogeneous data and also they are insensitive to the presence of erroneous or incomplete data presenting in the patients databases. General approach to the use of Bayesian networks and their recent successful application to some prognostic problems on the base of novel optimization technique are presented. The high-performance parallel method of the atomistic simulation of the ion beam sputtering deposition is applied to molecular dynamics modeling of the SiO2 thin films growth up to 100 nm containing several million atoms. It is possible to reveal dependence of film characteristics on the technological parameters.