A QSAR Model of hERG Binding Using a Large, Diverse, and Internally Consistent Training Set

Abstract

Over the past decade, the pharmaceutical industry has begun to address an addition to ADME/Tox profiling--the ability of a compound to bind to and inhibit the human ether-a-go-go-related gene (hERG)-encoded cardiac potassium channel. With the compilation of a large and diverse set of compounds measured in a single, consistent hERG channel inhibition assay, we recognized a unique opportunity to attempt to construct predictive QSAR models. Early efforts with classification models built from this training set were very encouraging. Here, we report a systematic evaluation of regression models based on neural network ensembles in conjunction with a variety of structure representations and feature selection algorithms. The combination of these modeling techniques (neural networks to capture non-linear relationships in the data, feature selection to prevent over-fitting, and aggregation to minimize model instability) was found to produce models with very good internal cross-validation statistics and good predictivity on external data.

A Composite Model for hERG Blockade

“…Many public and commercially available descriptor sets have been used to model hERG IC 50 values. Models based on 2D descriptors provided by MOE [40] and TSAR, [41] Ghose-Crippen, Kier & Hall topological indices, Isis-Keys, atom pair, electrotopological state descriptors, [42] molecular fingerprints, and molecular fragments have been published. [43][44][45][46] A large variety of statistical methods has also been used; self-organizing maps (SOMs), [39] multiple linear regression (MLR), partial least squares (PLS), logistic regression, [47] support vector machines (SVMs), Bayesian classifiers, and decision-tree algorithms.…”

In silicoADME/T modelling for rational drug design

“…For newly designed compounds, the homology model of the hERG channel can be used to perform docking or MD stimulation and assess their hERG toxicity according to their binding affinity. Seierstad and Agrafiotis developed a QSAR model for hERG toxicity prediction (Seierstad & Agrafiotis, 2006). This model can quantitatively assess the cardiotoxicity of newly designed compounds and provide alerts for compounds with a greater potential to cause toxicity.…”

In Silico Prediction of the Chemical Block of Human Ether-a-Go-Go-Related Gene (hERG) K+ Current

“…The ligand-based approach has been adopted to extract the features of hERG blockers in terms of inhibition of hERG current [9][10][11][12][13]. These methods include traditional two-dimensional (2D)-quantitative structure activity relationships (QSAR) [11,[14][15][16], 3D-QSAR [17][18][19][20][21][22], and classifi cation [23][24][25][26] methods. These methods extract chemical properties such as steric, hydrophobic, and electronic properties from the compounds.…”

“…Or a combinatorial method with a classification [23,24] based on a decision tree algorithm, which divides blockers into a hierarchy of homogeneous subgroups using the descriptors, can be used. By fi tting the compounds to subgroup-specifi c pharmacophore models, the predicted activities of compounds have been reported to become closer to their actual activities [9,16,46].…”