369
Chapter 12
Consensus Drug Design Using IT Microcosm
Pavel M. Vassiliev, Alexander A. Spasov, Vadim A. Kosolapov,
Aida F. Kucheryavenko, Nataliya A. Gurova and Vera A. Anisimova
© Springer Science+Business Media Dordrecht 2014
L. Gorb et al. (eds.), Application of Computational Techniques in Pharmacy and Medicine,
Challenges and Advances in Computational Chemistry and Physics 17,
DOI 10.1007/978-94-017-9257-8_12
P. M. Vassiliev () · A. A. Spasov · V. A. Kosolapov · A. F. Kucheryavenko · N. A. Gurova
Volgograd State Medical University (VSMU), Pavshikh Bortsov Sq. 1,
Volgograd 400131, Russian Federation
e-mail: pmv@avtlg.ru
V. A. Anisimova
Institute of Physical and Organic Chemistry at Southern Federal University (IPOC SFU),
Stachka Ave. 194/2, Rostov-on-Don 344090, Russian Federation
e-mail: anis39@mail.ru.
Abstract This chapter discusses Microcosm, an information technology package
for predicting the pharmacological activity of chemical compounds. This technology is based on a complex prediction methodology with a consensus approach to
prediction as its central component. The complex methodology of prediction in IT
Microcosm is essentially different from that of other QSAR approaches in that it
employs a redundant multi-descriptor, multi-level representation of the structure of
chemical compounds by an aggregate of fragment descriptors with different physicochemical meanings and varying extents of complexity. The methodology also
includes several classification methods that differ in their mathematical formalisms
and several decision making circuits that are conceptual in the results they yield. At
the same time, no feature space reductions are made, and no significant variables
are isolated; all of the parameters of description are used in the construction of the
prediction regularities. The integral decision rules are constructed by generalizing
the spectrum of primary prediction estimates using different levels and types of
consensus. In this chapter, we describe the paradigm of IT Microcosm, including
its theoretical concepts, a specialized QL language for chemical structure representation, and prediction methods and strategies using the package. The adequacy,
validity and high accuracy of IT Microcosm are demonstrated via sample predictions of the various pharmacological activities of structurally similar and structurally diverse organic compounds, complex organic salts, supramolecular complexes
and substance mixtures, accounting for the synergy between the individual components of mixtures. The authors also present the results of a successful application
of IT Microcosm, along with in vivo and in vitro experimental methods for (1) the
search for novel potent antioxidants, antiarrhythmics and antiplatelet agents; (2) the
optimization of the composition of supramolecular complexes with antioxidant and
Chapter 12
Consensus Drug Design Using IT Microcosm
Pavel M. Vassiliev, Alexander A. Spasov, Vadim A. Kosolapov,
Aida F. Kucheryavenko, Nataliya A. Gurova and Vera A. Anisimova
© Springer Science+Business Media Dordrecht 2014
L. Gorb et al. (eds.), Application of Computational Techniques in Pharmacy and Medicine,
Challenges and Advances in Computational Chemistry and Physics 17,
DOI 10.1007/978-94-017-9257-8_12
P. M. Vassiliev () · A. A. Spasov · V. A. Kosolapov · A. F. Kucheryavenko · N. A. Gurova
Volgograd State Medical University (VSMU), Pavshikh Bortsov Sq. 1,
Volgograd 400131, Russian Federation
e-mail: pmv@avtlg.ru
V. A. Anisimova
Institute of Physical and Organic Chemistry at Southern Federal University (IPOC SFU),
Stachka Ave. 194/2, Rostov-on-Don 344090, Russian Federation
e-mail: anis39@mail.ru.
Abstract This chapter discusses Microcosm, an information technology package
for predicting the pharmacological activity of chemical compounds. This technology is based on a complex prediction methodology with a consensus approach to
prediction as its central component. The complex methodology of prediction in IT
Microcosm is essentially different from that of other QSAR approaches in that it
employs a redundant multi-descriptor, multi-level representation of the structure of
chemical compounds by an aggregate of fragment descriptors with different physicochemical meanings and varying extents of complexity. The methodology also
includes several classification methods that differ in their mathematical formalisms
and several decision making circuits that are conceptual in the results they yield. At
the same time, no feature space reductions are made, and no significant variables
are isolated; all of the parameters of description are used in the construction of the
prediction regularities. The integral decision rules are constructed by generalizing
the spectrum of primary prediction estimates using different levels and types of
consensus. In this chapter, we describe the paradigm of IT Microcosm, including
its theoretical concepts, a specialized QL language for chemical structure representation, and prediction methods and strategies using the package. The adequacy,
validity and high accuracy of IT Microcosm are demonstrated via sample predictions of the various pharmacological activities of structurally similar and structurally diverse organic compounds, complex organic salts, supramolecular complexes
and substance mixtures, accounting for the synergy between the individual components of mixtures. The authors also present the results of a successful application
of IT Microcosm, along with in vivo and in vitro experimental methods for (1) the
search for novel potent antioxidants, antiarrhythmics and antiplatelet agents; (2) the
optimization of the composition of supramolecular complexes with antioxidant and