Acta Periodica Technologica 2017 Issue 48, Pages: 117-126
https://doi.org/10.2298/APT1748117K
Full text ( 682 KB)
Cited by
Chemometric and QSAR analysis of some thiadiazines as potential antifungal agents
Karadžić Milica Ž. (Faculty of Technology, Novi Sad)
Kovačević Strahinja Z. (Faculty of Technology, Novi Sad)
Jevrić Lidija R. (Faculty of Technology, Novi Sad)
Podunavac-Kuzmanović Sanja O. (Faculty of Technology, Novi Sad)
Quantitative structure-activity relationship (QSAR) analysis has been
performed in order to predict the antifungal activity of dihydroindeno and
indeno thiadiazines against toxigenic fungus Aspergillus flavus. The studied
compounds were classified according to their lipophilicity using the
principal component analysis (PCA). The partial least square regression
(PLSR) was used to distinguish the most important molecular descriptors for
non-linear modeling. Artificial neural networks (ANNs) were applied for the
antifungal activity prediction. The best QSAR models were validated by
statistical parameters and graphical methods. High agreement between the
observed and predicted antifungal activity values indicated the good quality
of the derived QSAR models. The obtained QSAR-ANN models can be used to
predict the antifungal activity of dihydroindeno and indeno thiadiazines and
of structurally similar compounds. The modeling of the antifungal activity
can contribute to the synthesis of new antifungal agents with better ability
to protect food and feed from the mycotoxins.
Keywords: artificial neural networks, mycotoxins, partial least square regression, QSAR, thiadiazines