QSAR modelling using combined simple competitive learning networks and RBF neural networks

The aim of this study was to propose a QSAR modelling approach based on the combination of simple competitive learning (SCL) networks with radial basis function (RBF) neural networks for predicting the biological activity of chemical compounds. The proposed QSAR method consisted of two phases. In the first phase, an SCL network was applied to determine the centres of an RBF neural network. In the second phase, the RBF neural network was used to predict the biological activity of various phenols and Rho kinase (ROCK) inhibitors. The predictive ability of the proposed QSAR models was evaluated and compared with other QSAR models using external validation. The results of this study showed that the proposed QSAR modelling approach leads to better performances than other models in predicting the biological activity of chemical compounds. This indicated the efficiency of simple competitive learning networks in determining the centres of RBF neural networks.     

Synthesis of carbon nanotube‐based nanocomposite and application for wastewater treatment by ultrasonicated adsorption process

The sorption of methylene blue (MB) and basic yellow 28 (BY28) dyes in water on Ag@ZnO/MWCNT (Ag‐doped ZnO loaded on multiwall carbon nanotubes) nanocomposite is investigated in a batch process, optimizing starting initial dye concentration, sonication time and adsorbent mass. Isotherms and kinetic behaviours of MB and BY28 adsorption onto Ag@ZnO/MWCNT were explained by extended Freundlich and pseudo‐second‐order kinetic models. Ag@ZnO/MWCNT was synthesized and characterized using X‐ray diffraction, energy‐dispersive X‐ray spectroscopy, field emission scanning electron microscopy and Brunauer–Emmett–Teller analysis. According to the experimental data, adaptive neuro‐fuzzy inference system (ANFIS), generalized regression neural network (GRNN), backpropagation neural network (BPNN), radial basic function neural network (RBFNN) and response surface methodology (RSM) were developed, and applied to forecast the removal performance of the sorbent. The influence of process variables (i.e. sonication time, initial dye concentration, adsorbent mass) on the removal of MB and BY28 was considered by central composite rotatable design of RSM, GRNN, ANFIS, BPNN and RBFNN. The performances of the developed ANFIS, GRNN, BPNN and RBFNN models were compared with RSM mathematical models in terms of the root mean square error, coefficient of determination, absolute average deviation and mean absolute error. The coefficients of determination calculated from the validation data for ANFIS, GRNN, BPNN, RBFNN and RSM models were 0.9999, 0.9997, 0.9883, 0.9898 and 0.9608 for MB and 0.9997, 0.9990, 0.9859, 0.9895 and 0.9593 for BY28 dye, respectively. The ANFIS model was found to be more precise compared to the other models. However, the GRNN method is much easier than the ANFIS method and needs less time for analysis. So, it has potential in chemometrics and it is feasible that the GRNN algorithm could be applied to model real systems. The monolayer adsorption capacity of MB and BY28 was 292.20 and 287.02 mg g^?1, respectively.     

蛋白质二级结构预测的人工神经网络方法研究

本文比较了五种神经网络方法预测蛋白质二级结构的准确率,并做出初步评价。五种神经网络分别是:误差反传前向网络(BP),径向基函数网络(RBF),广义回归神经网络(GRNN),串并联叠层网络(CF),Elman网络(ELM)。结果显示:GRNN的预测准确率达85.7%,优于其它网络。本文还讨论了训练集样本数及参数的优化对GRNN预测准确率的影响。     

基于偏最小二乘与广义回归神经网络的近红外光谱测定土豆中3种营养成分的研究

偏最小二乘(partial least squares,PLS)与广义回归神经网络(generalized regression neural networks,GRNN)联用对土豆样品建立起粗纤维、淀粉、蛋白质含量的预测校正模型,用PLS法将原始数据压缩为主成份,取前3个主成份的12个特征吸收峰输入GRNN网络,网络光滑因子σi为0.1.PLS-GRNN模型对样品3个组分含量的预测决定系数(R2)分别为: 0.945、 0.992、 0.938.结果表明,近红外光谱技术可以快速、准确地同时测定土豆中的粗纤维、淀粉、蛋白质,该方法可应用于果蔬产业的品质管理与控制.     

A Noncongeneric Model for Predicting Toxicity of Organic Molecules to Vibrio Fischeri

Abstract

A general Quantitative Structure-Activity Relationship (QSAR) model on Vibrio fischeri (Microtox^? test) was derived using the autocorrelation method for describing the molecules and a neural network as statistical tool. From a training set of 1068 organic chemicals described by means of four different autocorrelation vectors, it was possible to obtain valuable models but presenting some large outliers. Addition of the time of exposure as variable allowed us to derive a more powerful model from 2795 toxicity results. The predictive power of this 36/26/1 neural network model was tested on an external testing set of 385 toxicity data and compared with the performances of linear models designed for polar narcotic amines and for weak acid respiratory uncouplers.     

Pesticides as estrogen disruptors: QSAR for selective ERα and ERβ binding of pesticides

Evidence suggests that environmental exposure to estrogen-like compounds can cause adverse effects in humans and wildlife. The Endocrine Disruptor Screening and Testing Advisory Committee (EDSTAC) has advised screening of 87,000 compounds in the interest of human safety. This may best be accomplished by pre-screening using quantitative structure-activity relationship (QSAR) modelling. The present study aimed to develop in silico QSARs based on natural, semi-synthetic, synthetic, and phytoestrogens, to predict the potential estrogenic toxicity of pesticides. A diverse set of 170 compounds including steroidal-, synthetic- and phytoestrogens, as well as pesticides was used to construct the QSAR models using artificial neural networks (ANNs). Mean correlation coefficients between experimentally measured and predicted binding affinities were all greater than 0.7 and models had few false negative results, an important consideration for screening tools. This study demonstrated the utility of ANNs as QSAR models for pre-screening of potential endocrine disruptors.     

Artificial neural network model for the evaluation of chemical kinetics in thermally induced solid-state reaction

The artificial intelligence technique is utilized to improve evaluation of thermally induced solid-state reaction kinetics. A general regression neural network (GRNN) model was applied to directly determine the kinetic triplets, i.e., activation energy, pre-exponential factor, and mechanism model. The effect of number of heating rate on prediction performance of the GRNN model was assessed based on the estimation indictors. The obtained kinetic triplets based on the triple heating rates were considered to be accepted. The prediction ability of the GRNN model was very robust at more than three heating rates. The relative errors for kinetic parameters derived from five heating rates were within ±?4%, and the cognition rates for mechanism models were up to 99.6%. The developed GRNN model was successfully applied in the high-temperature synthesis of Li₄Ti₅O₁₂/C composites. It is expected that the model also could be extended to estimate the kinetics of other solid-state reactions.