COURSES

Goals

Getting acquainted with the computer methods to model the properties of molecules. The properties that are of interest in drug design and in chemical regulations - risk assessment of compounds in the environment will be addressed. Collecting knowledge about the databases that contain information about the structures and properties of compounds, and about the methods for data handling. Learning about the methods for encoding of molecular structures and for calculating molecular descriptors. Acquiring the knowledge about the statistical modeling and validation of models. Acquiring basic knowledge about molecular modelling – virtual scrambling, docking.

Curriculum

- Presentation of data banks, which are accessible via the Internet, and possibly compilation of students' own data banks for the different biological properties (dose and grades of toxicity, teratogenicity, carcinogenicity, binding constants of certain enzymes, etc.). - Encoding of chemical structures (SMILES notation, MDL, SDF, MOL). - Calculation of descriptors (topological, empirical, quantum-chemical, etc.) and use of relevant computer programs (DRAGON, CODESSA, etc.). - Use various software packages for the construction and validation of QSPR models (linear regression, principal component analysis, neural nets, etc.). - Modeling (linear and nonlinear): In this chapter, students will learn the basics of multiple linear regression (MLR) as an example of the linear regression. As example of the non-linear techniques various artificial neural networks will be presented. - Transformation of the measurement space: Some common measurement space transformation will be presented (e.g. PCA, etc.), which are used to enable the graphical presentation of the multidimensional metric space. - Clustering: We will present a simple procedure for clustering of data in a multidimensional space of measurements, as well as the use of artificial neural networks for the same purpose. - Model validation: We will learn the basic procedures for dividing data in different sets needed for model validation (learning and testing set). We will also discuss the methods used in various model validations. - Mathematical representation of chemical structures: Some simple representations of chemical structures that can be used in modeling the relationship between chemical structure and properties of molecules (QSAR, QSPR) will be discussed. - Practical application of the knowledge obtained – a case study of design of inhibitors of a selected enzyme.