Recently, research has been critically focused on finding new compounds with antirepellent
activity due to the rising of new types of mosquito-borne diseases. Mosquito
repellents are the safer and cleaner alternative to fight the anthropods from bitten
human skins, hence reduce the spread of diseases. This study investigated the
relationships between biological activity and structure of carboxamides by using
Quantitative Structure-Activity Relationship (QSAR) analysis. The data set used in this
study comprised of 40 carboxamide compounds taken from the literature with their
activities expressed as log PT (protection time). These compounds were split into
training set for model building and test set for external validation using activity-based
ranking method. The training set contained approximately 75% of the compounds
while the remaining compounds were then used as the validation set to verify the
accuracy of the model. DRAGON software was employed to generate molecular
descriptors. The important relevant descriptors were further selected and reduced by
using Genetic Algorithm (GA) as variable selection method. Two QSAR models were
developed by combining GA method with two different modelling techniques that are
multiple linear regressions (MLR) and partial least square (PLS). All the models are
robust with good correlation coefficient (r2) greater than 0.6 and external validation
r2test more than 0.5. Statistics of the GA-MLR model are r2 = 0.779 and r2test = 0.646.
Whereas, the second model generated from GA and PLS shows good r2 with value of
0.775 and r2test = 0.563. These results could be useful in finding new, safe, and more
effective repellents against Aedes Aegypti in a short time by providing guidance for
further laboratory work as well as prediction of external compounds and help to
understand the factors affecting their activity.