Title
In silico methods for classification of HIV-inhibitors
Poster
Description
Joanna Stoycheva 1 , Ismail Hdoufane2 , Alia Tadjer1 , Driss Cherqaoui2 , Jane Bogdanov3
1Faculty of Chemistry and Pharmacy, University of Sofia ‘‘St. Kliment Ohridski” 2 Department of Chemistry, Faculty of Sciences Semlalia, Marrakech, Moroc 3 Institute of Chemistry, Faculty of Natural Science and Mathematics, Ss. Cyril and Methodius University, Skopje, Macedonia
1Faculty of Chemistry and Pharmacy, University of Sofia ‘‘St. Kliment Ohridski” 2 Department of Chemistry, Faculty of Sciences Semlalia, Marrakech, Moroc 3 Institute of Chemistry, Faculty of Natural Science and Mathematics, Ss. Cyril and Methodius University, Skopje, Macedonia
Subject
1983 is the year when scientists isolated the retrovirus HIV, which uses reverse
transcriptase enzyme to replicate. Since then the search for inhibitors of this enzyme is
constantly increasing. Widely investigated are TIBO (Tetrahydroimidazo-[4,5,1-jk][1,4]-
benzodiazepinone) derivatives, which show high efficiency and less side effects. Object of
this study is finding the relationship between structure and biological activity of a series
TIBO derivatives and classifying the compounds into two groups - high and low activity.
Four different classifiers - support vector machines, artificial neural networks, random
forest and decision tree have been established. All classification models show good
classification rate ranges of the studied compounds and can be used in prediction of the
anti-HIV activities of untested molecules.
transcriptase enzyme to replicate. Since then the search for inhibitors of this enzyme is
constantly increasing. Widely investigated are TIBO (Tetrahydroimidazo-[4,5,1-jk][1,4]-
benzodiazepinone) derivatives, which show high efficiency and less side effects. Object of
this study is finding the relationship between structure and biological activity of a series
TIBO derivatives and classifying the compounds into two groups - high and low activity.
Four different classifiers - support vector machines, artificial neural networks, random
forest and decision tree have been established. All classification models show good
classification rate ranges of the studied compounds and can be used in prediction of the
anti-HIV activities of untested molecules.