Estimating the domain of applicability for machine learning QSAR models: a study on aqueous solubility of drug discovery molecules

We investigate the use of different Machine Learning methods to construct models for aqueous solubility. Models are based on about 4000 compounds, including an in-house set of 632 drug discovery molecules of Bayer Schering Pharma. For each method, we also consider an appropriate method to obtain error bars, in order to estimate the domain of applicability (DOA) for each model. Here, we investigate error bars from a Bayesian model (Gaussian Process (GP)), an ensemble based approach (Random Forest), and approaches based on the Mahalanobis distance to training data (for Support Vector Machine and Ridge Regression models). We evaluate all approaches in terms of their prediction accuracy (in cross-validation, and on an external validation set of 536 molecules) and in how far the individual error bars can faithfully represent the actual prediction error.     

Local lazy regression: making use of the neighborhood to improve QSAR predictions

Traditional quantitative structure-activity relationship (QSAR) models aim to capture global structure-activity trends present in a data set. In many situations, there may be groups of molecules which exhibit a specific set of features which relate to their activity or inactivity. Such a group of features can be said to represent a local structure-activity relationship. Traditional QSAR models may not recognize such local relationships. In this work, we investigate the use of local lazy regression (LLR), which obtains a prediction for a query molecule using its local neighborhood, rather than considering the whole data set. This modeling approach is especially useful for very large data sets because no a priori model need be built. We applied the technique to three biological data sets. In the first case, the root-mean-square error (RMSE) for an external prediction set was 0.94 log units versus 0.92 log units for the global model. However, LLR was able to characterize a specific group of anomalous molecules with much better accuracy (0.64 log units versus 0.70 log units for the global model). For the second data set, the LLR technique resulted in a decrease in RMSE from 0.36 log units to 0.31 log units for the external prediction set. In the third case, we obtained an RMSE of 2.01 log units versus 2.16 log units for the global model. In all cases, LLR led to a few observations being poorly predicted compared to the global model. We present an analysis of why this was observed and possible improvements to the local regression approach.     

Accurate solubility prediction with error bars for electrolytes: a machine learning approach

Accurate in silico models for predicting aqueous solubility are needed in drug design and discovery and many other areas of chemical research. We present a statistical modeling of aqueous solubility based on measured data, using a Gaussian Process nonlinear regression model (GPsol). We compare our results with those of 14 scientific studies and 6 commercial tools. This shows that the developed model achieves much higher accuracy than available commercial tools for the prediction of solubility of electrolytes. On top of the high accuracy, the proposed machine learning model also provides error bars for each individual prediction.     

支持向量机与KStar模型预测细胞色素P450酶催化的氧脱烃反应

分别以支持向量机(SVM)和KStar方法为基础, 构建了代谢产物的分子形状判别和代谢反应位点判别的嵌套预测模型. 分子形状判别模型是以272个分子为研究对象, 计算了包括分子拓扑、二维自相关、几何结构等在内的1280个分子描述符, 考查了支持向量机、决策树、贝叶斯网络、k最近邻这四种机器学习方法建立分类预测模型的准确性. 结果表明, 支持向量机优于其他方法, 此模型可用于预测分子能否被细胞色素P450酶催化发生氧脱烃反应. 代谢反应位点判别模型以538个氧脱烃反应代谢位点为研究对象, 计算了表征原子能量、价态、电荷等26个量子化学特征, 比较了决策树、贝叶斯网络、KStar、人工神经网络建模的准确率. 结果显示, KStar模型的准确率、敏感性、专一性均在90%以上, 对分子形状判别模型筛选出的分子, 此模型能较好地判断出哪个C―O键发生断裂. 本文以15个代谢反应明确的中药分子为验证集, 验证模型准确性, 研究结果表明基于SVM和KStar的嵌套预测模型具有一定的准确性, 有助于开展中药分子氧脱烃代谢产物的预测研究.     

便携式X射线荧光光谱法结合支持向量回归算法定量分析土壤中的砷含量

研究了基于统计学习理论的支持向量机（SVM）回归法在X射线荧光光谱定量分析中的应用。以39个农田土壤样品作为实验材料,以其中32个土壤样品作为校正集,选用SVM模型中Linear、Poly和RBF 3种核函数对As元素含量与荧光光谱数据进行回归建模。用3种不同模型对预测集中7个土壤样品的As元素含量进行预测分析,结果显示模型预测As元素含量与电感耦合等离子体发射光谱法测定的As元素含量之间的相关系数R2均大于0.99,相对分析误差RPD均大于3,表明所建立的SVM模型具有较好的使用价值。为了进一步考察SVM回归模型的预测效果,同应用较成熟的PLS回归模型的预测结果进行对比,结果显示SVM法的预测结果更好,表明SVM回归模型亦可用于便携式X射线荧光光谱法的定量预测分析。     

Three useful dimensions for domain applicability in QSAR models using random forest

One popular metric for estimating the accuracy of prospective quantitative structure-activity relationship (QSAR) predictions is based on the similarity of the compound being predicted to compounds in the training set from which the QSAR model was built. More recent work in the field has indicated that other parameters might be equally or more important than similarity. Here we make use of two additional parameters: the variation of prediction among random forest trees (less variation among trees indicates more accurate prediction) and the prediction itself (certain ranges of activity are intrinsically easier to predict than others). The accuracy of prediction for a QSAR model, as measured by the root-mean-square error, can be estimated by cross-validation on the training set at the time of model-building and stored as a three-dimensional array of bins. This is an obvious extension of the one-dimensional array of bins we previously proposed for similarity to the training set [Sheridan et al. J. Chem. Inf. Comput. Sci.2004, 44, 1912-1928]. We show that using these three parameters simultaneously adds much more discrimination in prediction accuracy than any single parameter. This approach can be applied to any QSAR method that produces an ensemble of models. We also show that the root-mean-square errors produced by cross-validation are predictive of root-mean-square errors of compounds tested after the model was built.     

机器学习方法用于二氢叶酸还原酶抑制剂的活性预测

分别采用支持向量学习机、人工神经网络、调节性逻辑回归和K-最临近等机器学习方法对761个二氢叶酸还原酶抑制剂建立了其活性分类预测模型. 采用组成描述符和拓扑描述符表征抑制剂的分子结构及物理化学性质, 使用Kennard-Stone方法进行训练集的设计, 并用Metropolis Monte Carlo模拟退火方法作变量选择. 结果表明, 支持向量学习机优于其它机器学习方法, 所得到的最优模型具有较好的预测结果, 其预测正确率为91.62%. 说明通过合适的训练集设计及变量选择, 支持向量学习机方法可以很好地用于二氢叶酸还原酶抑制剂的活性分类预测.