Acta Phys. -Chim. Sin. ›› 2009, Vol. 25 ›› Issue (04): 747-751.doi: 10.3866/PKU.WHXB200904281

• ARTICLE • Previous Articles     Next Articles

Prediction of Antitumor Activities of Indolo[1,2-b]Quinazoline Derivatives Using Electrotopological State Indices for AtomTypes

 MEI Hu, LIU Li, YANG Li, LI Jian, YAN Ning, WANG Qin   

  1. College of Bioengineering, Chongqing University, Chongqing 400044, P. R. China
  • Received:2008-10-12 Revised:2008-12-09 Published:2009-03-31
  • Contact: MEI Hu

Abstract: Electrotopological state indices for atomtype (ETSIAT)were employed to establish a quantitative structure- activity relationship (QSAR) model of antitumor activity for 17 indolo[1,2-b]quinazoline derivatives. Using step-wise regression analysis combined with the partial least squares (PLS) method, the coefficient of multiple determination R2, cross-validated coefficient of multiple determination Q2 (leave-one-out, LOO) and the root mean square error of estimation (RMSEE) of the optimal QSAR model were 0.806, 0.736, and 0.248, respectively. This optimal model was further validated by external validation. Results showed that 4 structural fragments, i.e., ≥N=, —NH—, =O, and >N— were closely correlated with the antitumor activities of indol[1,2-b]quinazoline derivatives. Furthermore, the structural fragment —NH— was negatively correlated with the antitumor activity while ≥N=, >N—, and =O were positively correlated with the antitumor activity. The substitution of R1 by strong electron-withdrawing groups may enhance compound antitumor activity and the steric effect at R2 may play an important role in the regulation of these activities. Based on the above observations, a total of 9 molecules were designed and predicted by using the optimal PLS model. Predicted activities of 4 molecules were 7.7%, 15.3%, 23.1%, and 130%higher than that of sample 13, respectively.

Key words: Electrotopological state indices for atomtype, Indolo[1,2-b]quinazoline derivative, Antitumor, Quantitative structure-activity relationship, Partial least squares model