CPSA与疏水性参数的遗传算法及神经网络法研究

Studies on the QSPR Between Charged Partial Surface Area and lgP of Organic Compounds Using Genetic Algorithm and Neural Network

用分子力学计算出“净”原子表面积 ,并利用量子化学方法计算出化合物的电荷加权部分表面积 (CP-SA) .在用遗传算法和神经网络法对改进的 CPSA与有机醇类化合物的疏水性参数作相关分析时发现 ,改进的 CPSA可有效地用于构效关系研究 ,且算法简洁易行 ,两种多元统计方法均得到了满意的结果 .

The original calculation methods of Charged Partial Surface Area(CPSA) are very complex because of considering many factors of environment, the calculation about the contribution in surface area of atom to molecular in these methods are solvent accessible area. In this paper, we improved the algorithm of CPSA and calculated the “nickel” surface area of atom using molecular mechanics and the CPSA of compounds by quantum chemistry. The satisfactory result of correlation analyze between lgP and improved CPSA of alcohol demonstrates that our algorithm is simple and effective, both genetic algorithm and neural network are well used in this research.