QSAR MODEL DEVELOPMENT FOR DNA BINDING CONSTANT OF COPPER COMPLEXES USING GA-MLR

Abstract

The ability of copper complexes to bind to DNA and exhibit antitumor and antimicrobial activity is well established. The
organic ligands of copper complexes do affect and regulate the activity by changing lipophilicity and intercalation to
DNA. Quantitative structure activity relationship (QSAR) analysis is widely applied drug design and development
method, here we make an attempt to develop a QSAR model for copper complexes, relating their physicochemical and
quantum chemical descriptors with DNA binding constant (Kb). A set of 30 copper complexes with reported Kb values
were selected from literature, these were built in Chemdraw and optimized using Semi-emprical PM3 method in
Hyperchem 7.5. Different physicochemical and quantum chemical descriptors were calculated for the complexes.
Double cross validation tool v2.0 was used to develop Multiple Linear Regression (MLR) model using Genetic Algorithm
(GA) based variable selection technique. Different QSAR models were developed using 20 complexes as training set and
the remaining 10 complexes were used as test set for external validation of the GA-MLR QSAR model. The best QSAR
model obtained showed R2 of 0.9076 and external validation Q2
test of 0.678. The statistically stable model thus obtained
gives us an opportunity to understand and develop better QSAR model based on advanced Quantum chemical protocols