Related Experiment Video
Updated: Jul 5, 2026
09:09
In Silico Clinical Trials for Cardiovascular Disease
Published on: May 27, 2022
Shape signatures: new descriptors for predicting cardiotoxicity in silico
Dmitriy S Chekmarev1, Vladyslav Kholodovych, Konstantin V Balakin
1Department of Pharmacology, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School and Environmental Bioinformatics and Computational Toxicology Center, 675 Hoes Lane, Piscataway, New Jersey 08854, USA.
Chemical Research in Toxicology
|May 9, 2008
Summary
Shape Signatures, a computational tool, effectively predicts cardiotoxicity by analyzing molecular shape and polarity. This method shows promise for identifying potential drug candidates with reduced cardiac risks.
Area of Science:
- Computational chemistry
- Toxicology
- Drug discovery
Background:
- Cardiotoxicity is a major concern in drug development.
- Predictive models are needed to assess drug safety early.
- Existing methods for cardiotoxicity prediction have limitations.
Purpose of the Study:
- To evaluate the Shape Signatures computational tool for predicting cardiotoxicity.
- To apply Shape Signatures to models for 5-HT(2B) receptor affinity and hERG channel inhibition.
- To assess the performance of Shape Signatures with k-NN, SVM, and SOM classification techniques.
Main Methods:
- Utilized a ray-tracing algorithm to generate molecular descriptors (Shape Signatures) encoding shape and polarity.
- Applied Shape Signatures descriptors to classification models for 5-HT(2B) and hERG interactions.
- Employed k-nearest neighbors (k-NN), support vector machines (SVM), and Kohonen self-organizing maps (SOM) for classification.
Main Results:
- Achieved 73-83% accuracy for 5-HT(2B) receptor computational models, comparable to commercial descriptors.
- Obtained 69-73% accuracy for hERG Shape Signatures-SVM models, consistent with existing computational approaches.
- Demonstrated that Shape Signatures descriptors perform well in heterogeneous property spaces.
Conclusions:
- Shape Signatures is a viable method for generating molecular descriptors for cardiotoxicity prediction.
- The methodology shows utility in classifying compounds interacting with 5-HT(2B) and hERG.
- These models can aid in predicting drug-induced cardiotoxicity during early drug discovery.
