Mostafa Zakariazadeh - Department of Biology, Faculty of Sciences, Payame Noor University, Tehran, Iran
Somaieh Soltani - Department of Medicinal Chemistry, Faculty of Pharmacy, Tabriz University of Medical Sciences, Tabriz, Iran

Nitroimidazoles compounds have been used in cancer therapy as radiosensitizers. They interact with DNAduplex structure via intercalating mode and change its conformer and interrupts its replication cycle. To theprediction of their radiosensitization effectiveness (pC۱.۶), a linear model was developed based on ۲D and۳D descriptors generated by PaDEL software. ۸۴ nitroimidazole compounds were divided into train (۶۲compounds) and test set (۲۲ compounds). The ۳D structures of compounds were prepared by Avogadrosoftware and geometry optimized by MMFF۹۴ force field. Descriptor selection was done using stepwiseregression. Developed linear models were validated based on the internal and external validations. ۹descriptors were selected from ۱۸۷۵ calculated descriptors. Selected descriptors were AlogP, BCUTc۱h,ASP۴, MLFER_A, MLFER_E, TDB۶s, TDB۱۰s, MOMI-X, and RDF۳۰m. The intercorrelatuon betweenselected descriptors were < ۰.۶۵. The correlatin coefficient between experimental and predicted pC۱.۶ were۰.۷۷ for the external test set. Obtained results indicated a good prediction potency for the developed modelin comparision to previous models. The selected descriptors represented the importance of both ۲-D and ۳-D characteristics of nitroimidazoles in their effect on DNA duplex conformation change. The obtained datacan be used for improve the effectiveness of studied compounds against cancer due to their binding to DNA.

QSAR; DNA-binding; nitroimidazole compounds, descriptor