Advances in Structure–Activity Relationship Studies on Potassium Channel Modulators - Ion Channels and Their Inhibitors - page 244

the functioning of ion channels, which in turn lead to aggrieved situations [ 17 ]. This
makes them as potential targets in drug discovery programs. However, because
of ion channels' ubiquitous nature and high variability even within a given family,
identification of drugs acting via them with specificity and high therapeutic value is
a challenge. This article discusses some of the recent advances made in quantitative
structure-activity relationship (QSAR) and modeling studies in the optimization of
various potassium channel modulators acting on different targets.

2 QSAR and Modeling Studies

Barring the accidental discoveries, identification and design of a ligand (or inhibi-
tor) for any given biological target is a complex process. Also, the drug research and
development is an interdisciplinary task of many areas, which include chemical,
biological, clinical, chemo- and bio-informatics. In this scenario, the quantitative
structure-activity relationship (QSAR) and modeling studies offer rationales to
make educated choice of biomolecular requirements by bridging the chemical
features and the activity phenomena.
The QSAR and modeling studies discussed in this article involved a variety
of physicochemical, quantum chemical, topological and topographical descriptors
from various sources such as Hansch and Leo's monograph [ 18 ], Molecular Orbital
Partial Atomic Charge (MOPAC) [ 19 ], ChemDraw's property/descriptor data-
base [ 20 ], Comprehensive Descriptors for Structural and Statistical Analysis
(CODESSA) [ 21 , 22 ] and DRAGON [ 23 ], for the parameterization of the chemical
structure. Most of the studies reviewed have involved 2D-QSAR approaches for the
rationalization of the biological function. They include simple Multiple Linear
Regression (MLR), Principle Components Regression (PCR), Partial Least Square
(PLS) [ 24 ], Genetic Function Approximation (GFA) [ 25 ], Genetic Partial Least
Squares (G/PLS) [ 26 ], Combinatorial Protocol in Multiple Linear Regression (CP-
MLR) [ 27 ], Discriminant and Cluster Analysis. Some of the reported works also
involved 3D-approaches in deriving the QSAR models. These approaches were
identified in the discussion of concerned models.
Most of the QSAR results presented in this article were validated through
various techniques such as leave one out, leave many out and external test sets.
For all regression equations, the statistics were reproduced as reported in the
original source. Briefly, in regression statistics n is the number of compounds in
the dataset, r is the correlation coefficient, Q 2 is cross-validated r 2 from leave-one/
many out procedure, s is the standard error of the estimate and F is the F-ratio
between the variances of calculated and observed activities. The values given in the
parentheses of regression equation are the standard errors (without arithmetic sign)
or 95% confidence limits (with arithmetic sign) of the regression coefficients.

Also for each study, wherever applicable, the QSAR methodology and the con-
tributing or modeling descriptors were highlighted and discussed to pull out the
essence of the investigation.