ABSTRACT
Computational techniques in drug discovery/development are routinely used as a versatile tool to minimize experiments and cost involved at each stage of the process. A large number of different computational aids/methods are currently applied in the rational drug discovery process. These include (quantitative) structure-activity/property/toxicity/pharmacokinetic relationship [(Q)SAR/QSPR/QSTR/QSPkR] models, molecular modeling, homology modeling, similarity searching, pharmacophores, drug discovery databases, machine learning, data mining, network analysis, and data analysis tools, etc. (Q)SAR/QSPR/QSTR/QSPkR approach conserves valuable resources and accelerates the drug development process by transforming searches for lead compounds into a mathematically quantified and computerized form using chemical intuition and experience. Molecular descriptors [MDs] are utilized to extract the structural information of the molecule in numerical form, suitable for model development and serve as a bridge between the molecular structure and physicochemical/biological activities of the molecules. Among various MDs used in (Q)SAR/QSPR/QSTR/QSPkR studies, graph-theoretical invariants or topological indices (TIs) are fundamental in nature and true structural invariants. Due to their conceptual simplicity, these MDs are easily computable using simple mathematical operations. They quantify certain aspects of molecular structure and are generally sensitive to different 454chemical features of the molecule such as size, shape, branching, symmetry, branching pattern, cyclicity, heterogeneity of atomic neighborhoods, etc. TIs can be classified as topostructural and topochemical depending upon their ability to distinguish among atoms or bond types present in the molecule. Based on their discriminating power, MDs are classified into six generations. The development of new TIs having high discriminating power but devoid of degeneracy continues to be a challenge in computational chemistry.
