433
Chapter 13
Continuous Molecular Fields Approach Applied
to Structure-Activity Modeling
Igor I. Baskin and Nelly I. Zhokhova
© 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_13
I. I. Baskin () · N. I. Zhokhova
Faculty of Physics, M. V. Lomonosov Moscow State University, Moscow, Russia 119991
e-mail: igbaskin@gmail.com
N. I. Zhokhova
e-mail: zhokhovann@gmail.com
Abstract The Method of Continuous Molecular Fields is a universal approach to
predict various properties of chemical compounds, in which molecules are represented by means of continuous fields (such as electrostatic, steric, electron density
functions, etc.). The essence of the proposed approach consists in performing statistical analysis of functional molecular data by means of joint application of kernel
machine learning methods and special kernels which compare molecules by computing overlap integrals of their molecular fields. This approach is an alternative
to traditional methods of building 3D “structure-activity” and “structure-property”
models based on the use of fixed sets of molecular descriptors. The methodology
of the approach is described in this chapter, followed by its application to building regression 3D-QSAR models and conducting virtual screening based on oneclass classification models. The main directions of the further development of this
approach are outlined at the end of the chapter.
13.1 Introduction
Currently, the leading role in predicting biological activity and physicochemical
properties of chemical compounds belongs to methods of chemoinformatics [1–3],
which are based on revealing “structure-activity” and “structure-property” relationships using modern statistical, data mining, machine learning and artificial intelligence approaches. Rapid progress of these techniques requires the development
of new tools of machine learning and data mining specially adapted to work with
molecular structures of chemical compounds [4].
In this section, we consider a new approach to building “structure-activity” and
“structure-property” models based on the use of continuous functions on space coordinates (called hereinafter continuous molecular fields) to represent molecular
Chapter 13
Continuous Molecular Fields Approach Applied
to Structure-Activity Modeling
Igor I. Baskin and Nelly I. Zhokhova
© 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_13
I. I. Baskin () · N. I. Zhokhova
Faculty of Physics, M. V. Lomonosov Moscow State University, Moscow, Russia 119991
e-mail: igbaskin@gmail.com
N. I. Zhokhova
e-mail: zhokhovann@gmail.com
Abstract The Method of Continuous Molecular Fields is a universal approach to
predict various properties of chemical compounds, in which molecules are represented by means of continuous fields (such as electrostatic, steric, electron density
functions, etc.). The essence of the proposed approach consists in performing statistical analysis of functional molecular data by means of joint application of kernel
machine learning methods and special kernels which compare molecules by computing overlap integrals of their molecular fields. This approach is an alternative
to traditional methods of building 3D “structure-activity” and “structure-property”
models based on the use of fixed sets of molecular descriptors. The methodology
of the approach is described in this chapter, followed by its application to building regression 3D-QSAR models and conducting virtual screening based on oneclass classification models. The main directions of the further development of this
approach are outlined at the end of the chapter.
13.1 Introduction
Currently, the leading role in predicting biological activity and physicochemical
properties of chemical compounds belongs to methods of chemoinformatics [1–3],
which are based on revealing “structure-activity” and “structure-property” relationships using modern statistical, data mining, machine learning and artificial intelligence approaches. Rapid progress of these techniques requires the development
of new tools of machine learning and data mining specially adapted to work with
molecular structures of chemical compounds [4].
In this section, we consider a new approach to building “structure-activity” and
“structure-property” models based on the use of continuous functions on space coordinates (called hereinafter continuous molecular fields) to represent molecular