Optimal descriptors calculated with the simplified molecular input line entry system (SMILES) have been utilized in modeling of carcinogenicity as continuous values (logTD50). These descriptors can be calculated using correlation weights of SMILES attributes calculated by the Monte Carlo method. A considerable subset of these attributes includes rare attributes. The use of these rare attributes can lead to overtraining. One can avoid the influence of the rare attributes if their correlation weights are fixed to zero. A function, limS, has been defined to identify rare attributes. The limS defines the minimum number of occurrences in the set of structures of the training (subtraining) set, to accept attributes as usable. If an attribute is present less than limS, it is considered “rare”, and thus not used. Two systems of building up models were examined: 1. classic training-test system; 2. balance of correlations for the subtraining and calibration sets (together, they are the original training set: the function of the calibration set is imitation of a preliminary test set). Three random splits into subtraining, calibration, and test sets were analysed. Comparison of abovementioned systems has shown that balance of correlations gives more robust prediction of the carcinogenicity for all three splits (split 1: rtest2=0.7514, stest=0.684; split 2: rtest2=0.7998, stest=0.600; split 3: rtest2=0.7192, stest=0.728).
展开▼
机译：通过简化的分子输入线输入系统（SMILES）计算出的最佳描述符已被用作致癌性建模的连续值（logTD 50 ）。可以使用通过蒙特卡洛方法计算的SMILES属性的相关权重来计算这些描述符。这些属性的相当一部分包括稀有属性。这些稀有属性的使用可能导致过度训练。如果将它们的相关权重固定为零，则可以避免稀有属性的影响。已定义函数limS来识别稀有属性。 limS定义训练（子训练）集合的结构集合中出现的最小次数，以接受可用属性。如果存在的属性小于limS，则将其视为“稀有”，因此不使用。研究了两个建立模型的系统：1.经典的训练测试系统; 2.子训练集和校准集的相关性平衡（它们都是原始训练集：校准集的功能是模仿初步测试集）。分析了三个随机拆分，分别是子训练，校准和测试集。上述系统的比较表明，相关性的平衡给出了所有三个分裂的致癌性的更可靠的预测（分裂1：r test 2 = 0.7514，s test = 0.684;拆分2：r test 2 = 0.7998，s test = 0.600;拆分3：r test 2 = 0.7192，s test = 0.728）。
展开▼
