Prediction of the logarithmic of partition coefficients (log P) of some organic compounds by least square-support vector machine (LS-SVM)

A new method least square-support vector machine (LS-SVM) was used to develop quantitative structure–property relationship (QSPR) models for predicting the logarithmic of n-octanol/water partition coefficient (log P) of some derivatives phenolic compounds. The calibration and predictive ability of LS-SVM were investigated and compared with those of three other methods; multiple linear regression (MLR), support vector linear regression (SVR) and artificial neural network (ANN). The results showed that the log P values calculated by LS-SVM were in good agreement with experimental values, and the performances of the LS-SVM models were comparable or superior to those of MLR, SVR and ANN methods. The root-mean-square errors of the training set and the predicting set for the LS-SVM model were 0.0855, 0.0746 and the squares of the correlation coefficients were 0.9960 and 0.9728, respectively. These values and other statistical parameters obtained for the LS-SVM model show the reliability of this model. LS-SVM is a new and effective method for predicting log P of some organic compounds, and can be used as a powerful chemometrics tool for QSPR studies.     

New group contribution method for the prediction of normal melting points

The melting point of organic compounds was estimated using a simple group contribution method. The optimum parameters of this new method were obtained using particle swarm optimization in a multivariate linear regression. The melting temperatures of 250 pure compounds were predicted, and the results were compared with experimental data and other models available in the literature. Compared to the currently used group contribution methods, the new method makes significant improvements in accuracy and applicability of this important property. The study shows that the proposed method presents an excellent alternative for the estimation of the melting temperature of organic compounds (AARD of 10%) from the knowledge of the molecular structure.     

Improved model for fullerene C<Subscript>60</Subscript> solubility in organic solvents based on quantum-chemical and topological descriptors

Fullerenes are sparingly soluble in many solvents. The dependence of fullerene’s solubility on molecular structure of the solvent must be understood in order to manage efficiently this class of compounds. To find such dependency ab initio quantum-chemical calculations in combination with quantitative structure–property relationship (QSPR) tool were used to model the solubility of fullerene C₆₀ in 122 organic solvents. A genetic algorithm and multiple regression analysis (GA-MLRA) were applied to generate correlation models. The best performance is accomplished by the four-variable MLRA model with prediction coefficient r _test² = 0.903. This study reveals a correlation of highest occupied molecular orbital energy (HOMO), certain heteroatom fragments, and geometrical parameters with solubility. Several other important parameters of solvents that affect the C₆₀ solubility have been also evaluated by the QSPR analysis. The employed GA-MLRA approach enhanced by application of quantum-chemical calculations yields reliable results, allowing one to build simple, interpretable models that can be used for predictions of C₆₀ solubility in various organic solvents.     

New concept of organic homo‐rank compounds and its application in estimating enthalpy of formation of mono‐substituted alkanes

Based on the topological characteristics of distance matrices and adjacency matrices of molecular graphs, a new concept of organic homo‐rank compounds was proposed. Based on this concept, compounds can be classified into new groups other than the traditional homologues. Furthermore, novel structure–property relationship approach named as homo‐rank compounds method can be developed. The feasibility of homo‐rank compounds method was explored by estimating the enthalpy of formation of organic compounds. The group contribution index (GCI_X) and group polarizability potential index (GPI_X) of substituents X were defined and determined for mono‐substituted alkanes RX (X includes 20 substituents). The research results show that the enthalpies of formation of organic homo‐rank compounds and their isomers can be correlated very well with the parameters GCI_X and GPI_X. Combining the method of homologues with that of homo‐rank compounds, a general and simple quantitative correlation equation (8) was established to estimate the enthalpy of formation for RX, and the calculation precision is within the chemical accuracy ‘1 kcal/mol’. For 242 samples of RX, the average absolute deviation between the experimental and the calculated values is 2.42 kJ/mol. In addition, the enthalpies of formation of more than 2800 samples of RX were estimated. The approaches of organic homo‐rank compounds and organic homologues are independent of but complementary to each other. The combination of these two methods can help us to understand the organic molecular structure–property relationships more deeply, and to investigate these relationships more conveniently and accurately. Copyright © 2015 John Wiley & Sons, Ltd.     

BP神经网络结合ATLD与三维荧光光谱法测量水中多环芳烃

多环芳烃(PAHs)是煤,石油,木材,烟草等燃料和有机高分子化合物等有机物不完全燃烧时产生的一种持久性有机污染物。迄今已发现有200多种PAHs,其中有多种PAHs具有致癌性。PAHs广泛分布于我们生活的环境中,水中的PAHs主要来源于生活污水,工业排水和大气沉降。使用三维荧光光谱法,结合BP神经网络与交替三线性分解(ATLD)算法对水中的PAHs进行定性和定量分析。以苊（ANA）和芴（FLU）2种PAHs为目标分析物,用甲醇（光谱级）制备样本。使用FS920稳态荧光光谱仪对样本进行检测,设置激发波长为200～370 nm,间隔10 nm记录一个数据;发射波长为240～390 nm,间隔2 nm记录一个数据。设置初始发射波长总是滞后激发波长40 nm,以消除一级瑞利散射的干扰。随后使用BP神经网络法对待测样本数据进行预处理。利用BP神经网络基于误差反向传播算法(error back propagation training,BP)原理,对测得的三维荧光数据进行数据压缩处理,该方法具有柔性的网络结构与很强的非线性映射能力,网络的输入层、隐含层和输出层的神经元个数可根据实际情况设定,并且网络的结构不同时,性能也有所差异。随后,用ATLD算法分解预处理后的三维荧光光谱数据。采用核一致诊断法确定待测样本的组分数为2。结果表明,ATLD算法分解得到两种PAHs(ANA和FLU)的激发、发射光谱图与目标光谱非常相似,能实现光谱重叠严重的PAHs(ANA和FLU)的快速定性和定量分析,实现了以“数学分离”代替“化学分离”。将预测样本导入训练好的BP神经网络中,得到处理后待测样本数据的网络均方差（MSE）均小于0.003,网络的峰值信噪比（PSNR）均大于120dB（数据压缩中典型的峰值信噪比值在30～40 dB之间,越高越好）,可见BP神经网络对样本数据的压缩效果较好。BP神经网络训练后,得到输出值与目标值之间的拟合度高,拟合系数达0.998,具有较好的数据压缩效果。使用ATLD算法对待测样本进行分解后得到平均回收率为97.1％和98.9％,预测均方根误差为0.081 8和0.098 5 μg·L-1。三维荧光光谱结合BP神经网络和ATLD能够实现痕量PAHs的快速检测。     

基于神经网络的选区激光熔化残余应力预测

当前对选区激光熔化产生的残余应力预测方法主要为数值模拟,但由于设备、环境、粉末等因素差异性较大,且具有较大不确定性,很难建立符合实际情况的数值模拟模型。利用神经网络在预测多变量、复杂线性信息处理方面能力强的特点,建立适用于预测316L不锈钢粉末选区激光熔化残余应力的模型。使用选区激光熔化技术打印相当数量的不同工艺参数的试样,采用超声波检测其内部残余应力作为神经网络的训练样本,并使用这些样本对神经网络模型进行训练,获得具有预测功能的神经网络,将验证样本的工艺参数输入神经网络,计算出预测的残余应力值,与实际检测值进行对比。实验结果表明,预测值与实际测量值偏差较小,验证了所提方法的有效性。采用神经网络预测残余应力的方法,可以快速确定不同选区激光熔化工艺参数对应的残余应力,避免设置残余应力较高的工艺参数,有效缩短制备高质量工件试样的周期,降低成本。     

链烷含氧衍生物的沸点和分子参数的QSPR研究

近年来分子的物理化学性质与其结构之间的相关性是热门研究领域之一,其中拓扑指数方法越来越受到重视并获得广泛应用. 本文用分子参数,即主原子数(N)、主链主原子数(N')、相对电负性(Xr)和取代参量(P)建立了一个定量预测链烷含氧衍生物沸点的数学模型.337个化合物计算值与相应的文献值都非常吻合,相关系数均大于0.9950,平均绝对误差小于3 ℃,平均相对误差小于1.5 %. 计算结果表明分子参数法的物理意义明确、操作简单,是进行定量构效相关(QSPR)研究的一种简便可行的方法.     

用分子子图法计算硝基呋咱化合物的生成热

用新的分子子图法计算硝基呋咱类化合物的生成热 ,将呋咱基团视为母体 ,即基子图项 ;硝基、叠氮基、甲基、氰基拆分为一个个原子 ,从原子的角度来分分子子图 ,将碳、氢、氧、氮原子视为取代基 ,即亚子图项 .同时考虑呋咱基团的个数 ,考虑 1位、2位、3位、4位上碳、氢、氧、氮原子及双键、叁键对生成热的影响 ,还考虑不饱和度、总硝基个数、环的个数 (除呋咱环 )、氮氮及氮氧双键的个数对生成热的影响 .用这种新的分子子图编码方法 ,对硝基呋咱化合物的生成热进行了拟合和预估 ,取得了满意的结果 ,其回归方程的相关系数达到了 0 .995 4 .     

Analysis of structure-Caco-2 permeability relationships using a property landscape approach

Understanding the relationship between the chemical structure of bioactive compounds and Caco-2 permeability is of major importance in modern drug discovery. The purpose of this work was to characterize systematically the Caco-2 permeability landscape of a benchmark dataset of 100 molecules using a novel approach based on the emerging concept of property landscape modeling. Pairwise comparisons of the Caco-2 permeability and chemical structures were calculated for all possible combinations in the dataset. To compare the chemical structures, two distinct manners to represent the molecules were employed, namely, continuous properties previously used to derive QSPR models and molecular fingerprints with different designs. We introduce the concept of “permeability cliffs” discussing cases of compounds with high molecular similarity but large permeability difference. All permeability cliffs were regarded as shallow cliffs, since no extreme difference in Caco-2 permeability (less than two log units) was identified in the dataset. A clear dependence of Caco-2 permeability landscape with molecular representation was observed. The current approach can be further extended to model other ADME relevant landscapes.     

Elman回归神经网络同时定量测定三种酚类化合物

应用Elman回归神经网络(ERNN)对光谱严重重叠的对-硝基苯酚,邻-硝基苯酚和2,4-二硝基苯酚体系的同时定量测定进行了研究,并与多变量线性回归(MLR)法作了比较。编制了PERNN和PMLR程序执行有关计算。通过最佳化确定了Elman回归网络的结构和参数。ERNN和MLR法所有组分的相对预测标准偏差(RSEP)分别为3.1%和2 027.3%,实验结果显示对于分辨严重重叠光谱本法是成功的。ERNN法是解决局部最小和提高收敛速度的一种有价值的工具,亦可用于分析全光谱而不只限于选取少数特征值。本法为不经预先分离同时测定严重重叠的分子光谱体系提供了新的途径。     

Simultaneous Determination of Multicomponents in Air Toxic Organic Compounds Using Artificial Neural Networks in Ftir Spectroscopy

The application of Artificial Neural Networks (ANNs) for nonlinear multivariate calibration using simulated FTIR data was demonstrated in this paper. Neural networks consisting of three layers of nodes were trained by using the back-propagation learning rule. Since parameters affect the performance of the network greatly, simulated data were used to train the network in order to get a satisfactory combination of all parameters. The mixtures of four air toxic organic compounds whose FTIR spectra are overlapped were chosen to evaluate the calibration and prediction ability of the network. The relative standard error (RSD%), the percent standard error of prediction samples (%SEP) and the percent standard error of calibration samples (%SEC) are used for evaluating the ability of the neural network.     

Quantitative structure–reactivity relationship studies on the catalyzed Michael addition reactions

Quantitative structure–reactivity relationship (QSRR) can be considered as a variant of quantitative structure–property relationship (QSPR) studies, where the chemical reactivity of reactants or catalysts in a specified chemical reaction is related to chemical structure. In this manner, the Michael addition of some different substrates using different catalysts (SDS, silica gel, and ZrOCl₂) was subjected to structure–reactivity relationship, quantitatively. Multiple linear regression (MLR) and partial least square (PLS) were used to perform the QSRR analysis. The resulted models for different catalyzed reactions showed that the catalysts probably act in different mechanisms since the models obtained for the catalysts included different parameters from substrate and enones. Overall, it was found that the reactivity in Michael addition reactions is controlled by coulombic (dipole and charge) interactions as well as the orbital energetic parameters. In the presence of different catalysts, the relative importance of these parameters is changed and hence the catalytic activity is changed. Copyright © 2008 John Wiley & Sons, Ltd.