Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Model selection based on structural similarity-method description and application to water solubility prediction.

Ralph Kühne1, Ralf-Uwe Ebert, Gerrit Schüürmann

  • 1Department of Ecological Chemistry, UFZ Centre for Environmental Research, Permoserstrasse 15, 04318 Leipzig, Germany.

Journal of Chemical Information and Modeling
|March 28, 2006
PubMed
Summary

This study introduces a novel method for selecting the best computational prediction models for chemical properties. It uses structural similarity and prediction errors to identify the most accurate scheme for specific compounds.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Prediction of the water solubility by a graph convolutional-based neural network on a highly curated dataset.

Journal of cheminformatics·2025
Same author

Assessment, distribution, and ecological risk of contaminants of emerging concern in a surface water-sediment-fish system impacted by wastewater.

The Science of the total environment·2024
Same author

In silico bioavailability triggers applied to direct and indirect thyroid hormone disruptors.

Chemosphere·2023
Same author

Improvement of binding pose prediction of the MR1 covalent ligands by inclusion of simple pharmacophore constraints and structural waters in the docking process.

3 Biotech·2023
Same author

Insight into the Mechanism Underlying <i>Dehalococcoides mccartyi</i> Strain CBDB1-Mediated B<sub>12</sub>-Dependent Aromatic Reductive Dehalogenation.

Environmental science & technology·2023
Same author

Amino Chemoassay Profiling of Aromatic Aldehydes-Unraveling Drivers of Their Skin Sensitization Potency.

Chemical research in toxicology·2023

Area of Science:

  • * Computational Chemistry
  • * Cheminformatics
  • * Quantitative Structure-Property Relationships (QSPR)

Background:

  • * Accurate prediction of chemical properties is crucial for drug discovery and materials science.
  • * Existing prediction methods often vary in performance depending on the compound class.
  • * A systematic approach is needed to select the most reliable prediction method for a given task.

Purpose of the Study:

  • * To develop and validate a method for selecting the optimal prediction scheme based on compound similarity.
  • * To improve the accuracy of property predictions by leveraging prediction errors on similar structures.
  • * To provide a framework for domain-specific optimization of predictive models.

Main Methods:

  • * Selection of structurally similar compounds using atom-centered fragments (ACFs) and k-nearest neighbor (kNN) analysis in 2D chemical space.

Related Experiment Videos

  • * Evaluation of prediction errors on these similar compounds to rank different property prediction methods.
  • * Application to seven water solubility estimation methods using a reference set of 1876 organic compounds.

    • Main Results:

    • * Demonstrated a reliable method for identifying the best-performing prediction scheme for a given compound and property.
    • * Showcased the effectiveness of using prediction errors on similar compounds for method selection.
    • * Compared the proposed approach with similarity-based error correction, highlighting its advantages.

    Conclusions:

    • * The developed method enhances the reliability of computational predictions by enabling informed selection of the best model.
    • * This approach offers a robust strategy for improving the performance of quantitative structure-property relationship (QSPR) models.
    • * The method is adaptable for specifying application domains, further refining prediction accuracy.