Replacement based non-linear data reduction in radial basis function networks QSAR modeling

Malek-Khatabi, Atefeh; Kompany‐Zareh, Mohsen; Gholami, Somayeh; Bagheri, Saeed

doi:10.1016/j.chemolab.2014.04.005

Cited by 13 publications

(4 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…QSAR related literature presents several methods used for the rational division of the dataset into training and test sets, such as: Based Activity Selection, Based k-means cluster, Sphere Exclusion Method, Principal Component Analysis, Self-Organizing Maps (SOM) and Kennard-Stone algorithm, among others [7][8][9][10][11][12]. The most commonly used in the QSAR studies are k-mean cluster, Based on Activity and Kennard-Stone algorithms.…”

Section: Introductionmentioning

confidence: 99%

Impact assessment of the rational selection of training and test sets on the predictive ability of QSAR models

Andrada

Vega-Hissi

Estrada

et al. 2017

SAR and QSAR in Environmental Research

View full text Add to dashboard Cite

This study performed an analysis of the influence of the training and test set rational selection on the quality and predictively of the quantitative structure-activity relationship (QSAR) model. The study was carried out on three different datasets of Influenza Neuraminidase (H1N1) inhibitors. The three datasets were divided into training and test sets using three rational selection methods: based on k-means, Kennard-Stone algorithm and Activity and the results were compared with Random selection. Then, a total of 31,490 mathematical models were developed and those models that presented a determination coefficient higher than: r > 0.8, r > 0.7, r > 0.5 and minimum standard deviation (SD) and minimum root-mean square error (RMS) were selected. The selected models were validated using the internal leave-one-out method and the predictive capacity was evaluated by the external test set. The results indicate that random selection could lead to erroneous results. In return, a rational selection allows for obtaining more reliable conclusions. The QSAR models with major predictive power were found using the k-means algorithm and selection by activity.

show abstract

Section: Introductionmentioning

confidence: 99%

Impact assessment of the rational selection of training and test sets on the predictive ability of QSAR models

Andrada

Vega-Hissi

Estrada

et al. 2017

SAR and QSAR in Environmental Research

View full text Add to dashboard Cite

show abstract

“…The performance of a RBFNN is greatly influenced by the number of RBF units (nh, the hidden layer centers). Too low nh gives rise to a poor estimation of relation even in the calibration set, and too many hidden layer neurons causes overfitting [26]. Therefore, the network parameters should be optimized before training.…”

Section: Methodsmentioning

confidence: 99%

Multi-Layer Identification of Highly-Potent ABCA1 Up-Regulators Targeting LXRβ Using Multiple QSAR Modeling, Structural Similarity Analysis, and Molecular Docking

et al. 2016

View full text Add to dashboard Cite

Abstract:In this study, in silico approaches, including multiple QSAR modeling, structural similarity analysis, and molecular docking, were applied to develop QSAR classification models as a fast screening tool for identifying highly-potent ABCA1 up-regulators targeting LXRβ based on a series of new flavonoids. Initially, four modeling approaches, including linear discriminant analysis, support vector machine, radial basis function neural network, and classification and regression trees, were applied to construct different QSAR classification models. The statistics results indicated that these four kinds of QSAR models were powerful tools for screening highly potent ABCA1 up-regulators. Then, a consensus QSAR model was developed by combining the predictions from these four models. To discover new ABCA1 up-regulators at maximum accuracy, the compounds in the ZINC database that fulfilled the requirement of structural similarity of 0.7 compared to known potent ABCA1 up-regulator were subjected to the consensus QSAR model, which led to the discovery of 50 compounds. Finally, they were docked into the LXRβ binding site to understand their role in up-regulating ABCA1 expression. The excellent binding modes and docking scores of 10 hit compounds suggested they were highly-potent ABCA1 up-regulators targeting LXRβ. Overall, this study provided an effective strategy to discover highly potent ABCA1 up-regulators.

show abstract

“…The theory of RBFNN has been adequately described in detail elsewhere [27][28][29]. So we will limit ourselves to a brief outline highlighting only the most important aspects.…”

Section: Radial Basis Function Neural Network (Rbfnns)mentioning

confidence: 99%

Application of unfolded principal component analysis–radial basis function neural network for determination of celecoxib in human serum by three-dimensional excitation–emission matrix fluorescence spectroscopy

Shahlaei

Bahrami

Abdolmaleki

et al. 2015

Spectrochimica Acta Part A: Molecular and Biomolecular Spectros

View full text Add to dashboard Cite

Replacement based non-linear data reduction in radial basis function networks QSAR modeling

Cited by 13 publications

References 42 publications

Impact assessment of the rational selection of training and test sets on the predictive ability of QSAR models

Impact assessment of the rational selection of training and test sets on the predictive ability of QSAR models

Multi-Layer Identification of Highly-Potent ABCA1 Up-Regulators Targeting LXRβ Using Multiple QSAR Modeling, Structural Similarity Analysis, and Molecular Docking

Application of unfolded principal component analysis–radial basis function neural network for determination of celecoxib in human serum by three-dimensional excitation–emission matrix fluorescence spectroscopy

Contact Info

Product

Resources

About