Electrotopological State Indices For Atom Type And Its Application To QSAR Study

As one of the most important methods in the modem theoretical and applied chemistry, Quantitative Structure-Activity Relationships (QSAR) studies have been widely used in drug designs and pollution evaluations. A reasonable method of structural description of a molecular is a key point in the QSAR research. As one of the most important 2-D molecular structural descriptors, electrotopological state indices can characterize the topological and electronic imformation of molecules. In this paper, electrotopological state indices for atom type (ETSIAT) was established and employed to the QSAR studies of toxicities or biologic activities of several datasets, and good results were obtained.ETSIAT was firstly employed in the quantitative structure-activity relationship (QSAR) studies on the aliphatic alcohol toxicity to the Tetrahymena pyriformis, tomato, spider, bacterial luciferases, Fathead minnow, and rat. The QSAR models were established by partial least squares and further validated by both internal and external validations. The results showed that ETSIAT can characterize the structural information relevant to aliphatic alcohol toxicity very well and the QSAR models established have good stabilities and prediction capabilities. From the PLS analysis on the QSAR models, it can be seen that hydrophobicity is the decisive factor to affect the aliphatic alcohol toxicity. The aliphatic alcohol toxicity was enhanced with increasing the length of carbon chain, and the aliphatic alcohol toxicity of straight chain was higher than that of branched chain with the equal number of carbon atoms. Due to the steric effect, there may be different toxicity mechanisms between long-chain and short-chain aliphatic alcohols.ETSIAT was then employed in the QSAR studies of Anti-Staphylococcurs aureus activities of 26 thiophene analogous, estrogen receptor binding affinity of 25 non-steroids compounds and antitumor activities of 17 indolo[1,2-b]quinazoline derivatives. Robust and predictive QSAR models were established. As for the study of the Anti-Staphylococcurs aureus activities of 26 thiophene analogous, by the method of step-wise regression followed by partial least squares, the coefficient of multiple determination R², cross-validated coefficient of multiple determination Q² (leave-one-out, LOO) and the Q²_ext (the multiple determination coefficient of external validation) of the resulting optimal QSAR model were 0.942, 0.701, and 0.941 respectively. The results showed that 4 structural fragments, i.e.-Cl, -F, -OH and benzene ring, were closely correlated with the antibacterial activities of thiophene analogous. All of the 4 structural fragments were positively correlated with the antibacterial activities. Besides, the strong electron-withdrawing substituent of benzene ring may enhance the antibacterial activities of the samples. As for the estrogen receptor binding affinity of 25 non-steroids compounds, the R², Q² (LOO) and the Q²_ext of the optimal QSAR model were 0.722, 0.662 and 0.53 respectively. The results showed that 2 structural fragments, i.e. -OH and -CH3, were closely correlated with the binding affinity of non-steroids compounds. And the structural fragment–OH was positively correlated with the binding affinity, while -CH3 was negatively correlated with the binding affinity. Besides, the hydroxyl on the substituent may enhance the binding affinity of the samples. And lesser terminal C atoms of the sample may also benefit the binding affinity. As for antitumor activities of 17 indolo[1,2-b]quinazoline derivatives, Using step-wise regression analysis combined by partial least squares, the R², Q² (LOO), and Q²_ext of the optimal QSAR model were 0.806, 0.736, and 0.846 respectively. The results showed that 4 structural fragments, i.e.≥N=, -N-, =O, and >N-, were closely correlated with the antitumor activities of indolo[1,2-b]quinazoline derivatives. Furthermore,the structural fragment -N- was negatively correlated with the antitumor activity, while≥N=,>N- and =O were positively correlated with the antitumor activity. Besides, the substitution of R1 by strong electron-withdrawing groups may enhance the antitumor activities of the compounds...