Drug development is a long and time-consuming process, which can take an
average time of 10 years for the identification of one lead compound to be further tested
in the preclinical phases. Quantitative Structure-Activity Relationships (QSAR)-based
techniques are valuable tools for shortening the time of lead compound identification,
but also for focusing and limiting the time-costly synthetic activities and
biological/ADMET evaluations. This review reports an overview of the current research
and potential applications of QSAR modelling tools in rational drug design. The chapter
is set out in the same order in which QSAR models are generally built up, starting from
the setup of the dataset for modelling, assembly of typical molecular descriptors and
selection routes, followed on by an outline of the commonly used techniques for
establishing the QSAR models and lastly, by a discussion of the most useful procedures
for reliability and uncertainty assessment of the models for regulatory purposes.
Keywords: Rational drug design, QSAR modelling, Linear discriminant analysis,
Principal component regression, Partial least squares regression, Decision tree and
decision forest, Artificial neural networks, Support vector machines, Gene
expression programming theory, QSAR reliability, QSAR validation, QSAR
applicability domain.
