SVM-Based Feature Selection for Characterization of Focused Compound Collections

Byvatov, Evgeny; Schneider, Gisbert

doi:10.1021/ci0342876

Cited by 63 publications

(53 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The original idea for the use of the local gradient came from previous pioneering studies. [16][17][18] The local gradient was calculated using the scripts written in our laboratory in the R environment [19] In a manner similar to the Bayesian analysis, the value of the local gradient of each substructure was used as the basis for the atom score. [20,21] The value of the local gradient of each ECFP substructure is divided by the number of heavy atoms present in the substructure, and the calculated score value is assigned to each atom.…”

Section: Atom Colouringsmentioning

confidence: 99%

Visualization of Molecular Selectivity and Structure Generation for Selective Dopamine Inhibitors

Hasegawa¹,

Migita²,

Funatsu³

2010

Molecular Informatics

View full text Add to dashboard Cite

Activity landscapes were used in combination with atom colourings for the visualization of molecular selectivity. Multiple inhibitory activities in the dopamine family were selected in order to derive its molecular selectivity. All molecular structures were mapped in 2D chemical space by preserving the relative distance between any pair of molecules using multidimensional scaling. The values for the inhibitory activity against each dopamine isoenzyme (D2, D3, and D4) were added independently to the data points in 2D chemical space. Three activity landscapes were generated after carrying out colour-graded interpolation between the data points. The activity landscapes facilitated the detection of three specific active regions and the corresponding specific inhibitors for D2, D3, and D4 isoenzymes. Three support vector regression models were separately constructed using extended connectivity fingerprint descriptors and each inhibitory data set. By applying the atom scores along with the graded-colours to the specific inhibitors, the molecular selectivity differentiating each dopamine isoenzyme could be understood visually. Furthermore, the landscape technique was combined with structure generation to produce chemical structures that stay within the D3 specific active region.

show abstract

Section: Atom Colouringsmentioning

confidence: 99%

Visualization of Molecular Selectivity and Structure Generation for Selective Dopamine Inhibitors

Hasegawa¹,

Migita²,

Funatsu³

2010

Molecular Informatics

View full text Add to dashboard Cite

show abstract

“…There has been much research effort put into the field of feature extraction. 59 The problem of selecting properties which are responsible for given outputs occurs in various machine learning applications. [60][61][62] We use feature selection methods with the objective to detect features that are responsible for the underlying class structure.…”

Section: Machine Learning Techniquesmentioning

confidence: 99%

Classification of Highly Unbalanced CYP450 Data of Drugs Using Cost Sensitive Machine Learning Techniques.

Eitrich¹,

Kless²,

Druska³

et al. 2007

ChemInform

View full text Add to dashboard Cite

In this paper, we study the classifications of unbalanced data sets of drugs. As an example we chose a data set of 2D6 inhibitors of cytochrome P450. The human cytochrome P450 2D6 isoform plays a key role in the metabolism of many drugs in the preclinical drug discovery process. We have collected a data set from annotated public data and calculated physicochemical properties with chemoinformatics methods. On top of this data, we have built classifiers based on machine learning methods. Data sets with different class distributions lead to the effect that conventional machine learning methods are biased toward the larger class. To overcome this problem and to obtain sensitive but also accurate classifiers we combine machine learning and feature selection methods with techniques addressing the problem of unbalanced classification, such as oversampling and threshold moving. We have used our own implementation of a support vector machine algorithm as well as the maximum entropy method. Our feature selection is based on the unsupervised McCabe method. The classification results from our test set are compared structurally with compounds from the training set. We show that the applied algorithms enable the effective high throughput in silico classification of potential drug candidates.

show abstract

“…The usefulness of SVMs in drug design has for instance recently been shown by Byvatov et al [4,5]. In SVMs one usually chooses W[f] 1 ³2 k w k 2 where w is the weight vector of the separating hyperplane in feature space, and k¥ k denotes the Euclidian norm.…”

Section: Descriptor Selection and Machine Learningmentioning

confidence: 99%

Towards Optimal Descriptor Subset Selection with Support Vector Machines in Classification and Regression

2004

View full text Add to dashboard Cite

In this paper we present a novel method for selecting descriptor subsets by means of Support Vector Machines in classification and regression ± the Incremental Regularized Risk Minimization (IRRM) algorithm. In contrast to many other wrapper methods it is fully deterministic and computationally efficient. We compare our method to existing algorithms and present results on a Human Intestinal Absorption (HIA) classification data set and the Huuskonen regression data set for aqueous solubility.

show abstract

SVM-Based Feature Selection for Characterization of Focused Compound Collections

Cited by 63 publications

References 28 publications

Visualization of Molecular Selectivity and Structure Generation for Selective Dopamine Inhibitors

Visualization of Molecular Selectivity and Structure Generation for Selective Dopamine Inhibitors

Classification of Highly Unbalanced CYP450 Data of Drugs Using Cost Sensitive Machine Learning Techniques.

Towards Optimal Descriptor Subset Selection with Support Vector Machines in Classification and Regression

Contact Info

Product

Resources

About