2.10.  Principal component analysis (PCA)
The process of drug discovery has come to involve a critical concept known as ‘Chemical space’ which is defined as a multidimensional space projection of the number of property descriptors calculated for each chemical entity. PCA is a method to visualize chemical space in lower dimensions in order to identify and underline dominant patterns of drug like entities. The term PCA was first coined by Karl Pearson in 1901 and is an application of linear algebra (Ahmad, 2019, Khan et al., 2018).
Osiris Property Explorer 4.5.1 was used for defining and visualizing multivariate datasets of prospective drug candidates from WS and standard reference drugs through comparison of properties like TPSA, percent absorption, MW, hydrogen bond donor, hydrogen bond acceptor, number of rotatable bonds, Lipinski’s violations, leadlikeness and BAS. PCA helps in reducing the dimensionality of the dataset and increases interpretability. It does so by creating new uncorrelated variables which maximize the variance successively. Another added advantage of PCA is a 3D visualization in chemical space of how ‘drug-like’ are the molecules under study to known standard drugs in terms of their proximity to them in 3D chemical space.