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.     

Artificial Neural Network (ANN) design for Hg–Se interactions and their effect on reduction of Hg uptake by radish plant

The tendency of selenium to interact with heavy metals in presence of naturally occurring species has been exploited for the development of green bioremediation of toxic metals from soil using Artificial Neural Network (ANN) modeling. The cross validation of the data for the reduction in uptake of Hg(II) ions in the plant R. sativus grown in soil and sand culture in presence of selenium has been used for ANN modeling. ANN model based on the combination of back propagation and principal component analysis was able to predict the reduction in Hg uptake with a sigmoid axon transfer function. The data of fifty laboratory experimental sets were used for structuring single layer ANN model. Series of experiments resulted into the performance evaluation based on considering 20% data for testing and 20% data for cross validation at 1,500 Epoch with 0.70 momentums The Levenberg–Marquardt algorithm (LMA) was found as the best of BP algorithms with a minimum mean squared error at the eighth place of the decimal for training (MSE) and cross validation.     

机器学习方法用于选择性环氧化酶-2抑制剂活性预测模型的建立

与传统的非甾体类消炎药相比,选择性环氧化酶-2抑制剂具有无胃肠道粘膜损伤,溃疡和肾功能障碍等严重的副作用,设计选择性环氧化酶-2抑制剂具有重要意义。本文用支持矢量学习机和神经网络两种机器学习方法建立选择性环氧化酶-2抑制剂的活性预测模型,以期为选择性环氧化酶-2抑制剂药物的合成提供先导化合物。我们将467个环氧化酶-2抑制剂用Kennard-Stone方法分为训练集,验证集和独立测试集,对每一抑制剂分子我们计算了463个包含组成描述符和拓扑描述符的分子描述符来表征其分子结构,并通过F-Score方法选取最重要的分子描述符用于分类模型的建立。结果表明,SVM方法通过变量筛选后具有很好的预测能力,其预测正确率达到93.30%。     

人工神经网络预测非环状亚硝胺急性毒性的QSAR研究

在DFT-B3LYP/6-311+G(d,p)水平对60种非环状亚硝胺分子结构进行几何全优化,通过多元逐步线性回归(MSR)分析筛选出9个量子化学描述符作为自变量,log LD50(lethal dose 50％,LD50:大鼠口服急性毒性)作为因变量,采用人工神经网络(ANN)方法构建QSAR模型.经Levenber...     

Neural network modelling of an electrochemical process

Artificial Neural Network (ANN) methodology has gained popularity in chemistry in recent years as a result of its ability to solve problems for various purposes. In particular, ANN shows a very high performance in the modelling of the experimental measurements. To this aim, the trained ANN is used to predict unknown values of measurement system. For a given trial set of parameters, the experimental response may be predicted by the model. In recent decades, several investigations were based on the electrooxidation of D-glucose owing to its many applications such as detection systems (glucose sensors), fuel cells and synthesis of economically interesting products. In the present work based on the electrochemical process modelling, the estimation of peak current densities of the D-glucose electrooxidation on palladium electrode in alkaline medium was investigated as a function of potential sweep rate and D-glucose concentration by using a three layer feed-forward ANN with error propagation learning algorithm.     

Calcium mediated polyelectrolyte adsorption on like-charged surfaces

Monte Carlo simulations within the primitive model of calcium-mediated adsorption of linear and comb polyelectrolytes onto like-charged surfaces are described, focusing on the effect of calcium and polyion concentrations as well as on the ion pairing between polymers and calcium ions. We use a combination of Monte Carlo simulations and experimental data from titration and calcium binding to quantify the ion pairing. The polymer adsorption is shown to occur as a result of surface overcharging by Ca(2+) and ion pairing between charged monomers and Ca(2+). In agreement with experimental observations, the simulations predict that the polymer adsorption isotherm goes through a maximum as the calcium or the polymer concentration is increased. The non-Langmuir isotherms are rationalized in terms of charge-charge correlations.