Quantitative Structure-Activity Relationships (QSARs) are models relating variation of molecule properties, such as biological activities, to variation of some structural features of chemical compounds. Three main topics take part of the QSAR/QSPR approach to the scientific research: the representation of molecular structure, the definition of molecular descriptors and the chemoinformatics tools. Molecular descriptors are numerical indices encoding some information related to the molecular structure. They can be both experimental physico-chemical properties of molecules and theoretical indices calculated by mathematical formulas or computational algorithms. In the last few decades, much interest has been addressed to studying how to encompass and convert the information encoded in the molecular structure into one or more numbers used to establish quantitative relationships between structures and properties, biological activities or other experimental properties. Autocorrelation descriptors are a class of molecular descriptors based on the statistical concept of spatial autocorrelation applied to the molecular structure. The objective of this chapter is to investigate the chemical information encompassed by autocorrelation descriptors and elucidate their role in QSAR and drug design. After a short introduction to molecular descriptors from a historical point of view, the chapter will focus on reviewing the different types of autocorrelation descriptors proposed in the literature so far. Then, some methodological topics related to multivariate data analysis will be overviewed paying particular attention to analysis of similarity/diversity of chemical spaces and feature selection for multiple linear regressions. The last part of the chapter will deal with application of autocorrelation descriptors to study similarity relationships of a set of flavonoids and establish QSARs for predicting affinity constants, Ki, to the GABA A benzodiazepine receptor site, BzR.
Consonni, V., Todeschini, R. (2011). Structure-Activity Relationships by autocorrelation descriptors and genetic algorithms. In H. Lodhi, Y. Yamanishi (a cura di), Chemoinformatics and Advanced Machine Learning Perspectives: Complex Computational Methods and Collaborative Techniques (pp. 60-93). Hershey (PA) : IGI Global [10.4018/978-1-61520-911-8.ch005].
Structure-Activity Relationships by autocorrelation descriptors and genetic algorithms
CONSONNI, VIVIANA
;TODESCHINI, ROBERTO
2011
Abstract
Quantitative Structure-Activity Relationships (QSARs) are models relating variation of molecule properties, such as biological activities, to variation of some structural features of chemical compounds. Three main topics take part of the QSAR/QSPR approach to the scientific research: the representation of molecular structure, the definition of molecular descriptors and the chemoinformatics tools. Molecular descriptors are numerical indices encoding some information related to the molecular structure. They can be both experimental physico-chemical properties of molecules and theoretical indices calculated by mathematical formulas or computational algorithms. In the last few decades, much interest has been addressed to studying how to encompass and convert the information encoded in the molecular structure into one or more numbers used to establish quantitative relationships between structures and properties, biological activities or other experimental properties. Autocorrelation descriptors are a class of molecular descriptors based on the statistical concept of spatial autocorrelation applied to the molecular structure. The objective of this chapter is to investigate the chemical information encompassed by autocorrelation descriptors and elucidate their role in QSAR and drug design. After a short introduction to molecular descriptors from a historical point of view, the chapter will focus on reviewing the different types of autocorrelation descriptors proposed in the literature so far. Then, some methodological topics related to multivariate data analysis will be overviewed paying particular attention to analysis of similarity/diversity of chemical spaces and feature selection for multiple linear regressions. The last part of the chapter will deal with application of autocorrelation descriptors to study similarity relationships of a set of flavonoids and establish QSARs for predicting affinity constants, Ki, to the GABA A benzodiazepine receptor site, BzR.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
