Chemoinformatics and Molecular Modeling Laboratory was established in 2014. The main purpose of our laboratory is to use data mining tools and variety of molecular modeling technologies for solving practical problems of chemistry.
The following areas are studied in Chemoinformatics and Molecular Modeling Laboratory:
using and development of chemoinformatics and molecular modeling tools for drug design and design of new materials;
modeling of organic and metabolic reactions by methods of chemoinformatics: from empirical to predictive chemistry;
data mining for translational medicine.
Research work: Using and development of chemoinformatics and molecular modeling tools for drug design and design of new materials
The aim of the project: using data mining tools and the variety of molecular modeling technologies for solving practical problems of chemistry.
The task of the project: investigation of the nature of the chemical bonds and the reactivity of the compounds by means of quantum chemistry tools, development of non-classical analysis approaches “structure-property”, including capable of working with complex chemical entities (such as mixtures of compounds, solutions and materials), computer-aided design of new drugs on the basis of ligand and biological target structure.
Methods and Approaches:
development of the methodology of the continuous fields method for three-dimensional “structure-property” modeling (3DQSAR);
models development for the simulation of the properties which are important in terms of antimalarial activity;
development of the methodology for modeling of molecular properties in solutions for reactions of bimolecular substitution;
study the nature of intermolecular bonds by the approaches from “Atoms in Molecules” theory and “interacting quantum atom” approach.
The results of these researches have fundamental importance for modern organic chemistry and are found to be of great interest for chemists, theoreticians and experimentalists. The methodology of modeling of the properties of the compounds in solutions has been developed and tested, nature of binding of selenium atom with IIIA group acceptors which are important in terms of creation of new materials for semiconductors has been studied, the method “3DQSAR” has been developed - these are our theoretical results in the framework of this project. The model that allows predicting the redox potential of organic molecules (it is an important descriptor of antimalarial activity) has been obtained and made available - this is the practical result.
Research work: Chemoinformatics approaches to organic and metabolic reactions: from empirical to predictive chemistry
This project is carried out in the framework of the project of Russian Science Foundation 14-43-00024.
The aim of the project: the development of data mining methods for chemical reactions, including predictive modeling of parameters of chemical reactions and their conditions and analysis of large databases.
The task of the project: the development of the common methodology for data mining approach for reactions on the basis of the condensed graph of a reaction (CGR), which allows using multiple cheminformatics approaches designed for individual molecules on reactions data.
Methods and Approaches:
representation of complex reactions (including uncompleted and multi-step) by CGR;
development of property enriched fragment descriptors for reactions;
reaction similarity approach based on CGR;
inductive Learning Transfer methods for improving predictive performance of QSRR models;
method of visualization and analysis of large chemical reaction spaces.
The research results are fundamentally important for the modern organic chemistry and are found to be of great interest for chemists, theoreticians and experimentalists. The theorists will have new approaches and algorithms for chemical reactions data mining, and experimenters will have access through the Internet to the service with a user-friendly interface, which will help get answers to questions arising in the synthesis planning.
Research work Data mining for translational medicine¶
This project aims at the development and implementation of integrated computational methods for the reprofiling of drugs, using a variety of online available data sets. These arrays include unstructured texts (social media networking, biomedical literature, and electronic medical records) and electronic databases of the biological interactions of chemical compounds (including a biological screening of large chemical libraries). Furthermore, the translation of data into a format suitable for quantitative modeling of drug effect will allow integrating the text and laboratory data for creation a large amount of meta-data, suitable for building quantitative models linking the chemical structure and biological activity.
The proposed project links together the methods for the analysis of text data in information networks, the results of the study of biological networks, modeling the interaction of chemical drugs with their biological targets, and analysis of specialized biomedical literature and electronic data obtained from patients. The proposed computer system uniquely integrates a range of data (and related techniques) from person to molecules and again to the person in order to establish a previously unknown drug-protein and drug-disease bonds.