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

Ensemble methods for classification in cheminformatics.

Christian Merkwirth1, Harald Mauser, Tanja Schulz-Gasch

  • 1Computational Biology & Applied Algorithmics Group, Max-Planck-Institut für Informatik, Stuhlsatzenhauseg 85, 66123 Saarbrücken, Germany, and Roche Pharma Research, Basel, Switzerland. cmerk@mpi-sb.mpg.de

Journal of Chemical Information and Computer Sciences
|November 24, 2004
PubMed
Summary
This summary is machine-generated.

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

No differences in clinical and radiographic outcomes between standard and shortened versions of a same cementless femoral stem in Dorr type A femurs.

Hip international : the journal of clinical and experimental research on hip pathology and therapy·2026
Same author

Single-stage revision for chronic fistulized infected total knee arthroplasty achieves infection-free survival comparable to non-fistulized cases.

Knee surgery, sports traumatology, arthroscopy : official journal of the ESSKA·2026
Same author

Geno2pheno: recombination detection for HIV-1 and HEV subtypes.

NAR molecular medicine·2025
Same author

Shear-induced bubble nucleation in magmas.

Science (New York, N.Y.)·2025
Same author

The "five-star matrix" for patient-centric drug discovery and development in neuroscience.

Neuron·2025
Same author

Biological databases in the age of generative artificial intelligence.

Bioinformatics advances·2025
Same journal

Future Papers.

Journal of chemical information and computer sciences·2016
Same journal

Future Papers.

Journal of chemical information and computer sciences·2016
Same journal

Future Papers.

Journal of chemical information and computer sciences·2016
Same journal

Future Papers.

Journal of chemical information and computer sciences·2016
Same journal

Future Papers.

Journal of chemical information and computer sciences·2016
Same journal

Future Papers.

Journal of chemical information and computer sciences·2016
See all related articles

Ensemble methods, including k-nearest neighbors and support vector machines (SVMs), achieved over 90% accuracy in classifying pharmaceutical compounds. These robust models also halved misclassification rates for predicting unspecific protein inhibition.

Area of Science:

  • Computational chemistry
  • Cheminformatics
  • Machine learning in drug discovery

Background:

  • Binary classification is crucial for drug discovery, aiding in predicting compound properties.
  • Ensemble methods offer potential for improved predictive performance in complex datasets.

Purpose of the Study:

  • To apply and compare ensemble methods for binary classification of pharmaceutical compounds.
  • To evaluate model robustness with more features than observations.
  • To enhance prediction accuracy for drug-like substance properties.

Main Methods:

  • Application of ensemble and single models: k-nearest neighbors, support vector machines (SVMs), and ridge regression.
  • Comparison of various model variants on two distinct pharmaceutical compound datasets.

Related Experiment Videos

  • Robust classification analysis, particularly in high-dimensional scenarios (more features than observations).

    • Main Results:

    • Classification rates exceeding 90% were achieved for both cytochrome P450 inhibition and Frequent Hitters datasets.
    • Ensemble models demonstrated robust performance even when feature numbers exceeded sample sizes.
    • A twofold reduction in cross-validated misclassification rate was observed for the Frequent Hitters problem compared to prior studies.

    Conclusions:

    • Ensemble methods provide a powerful and robust approach for binary classification in pharmaceutical research.
    • These methods significantly improve prediction accuracy for critical drug-like substance properties.
    • The findings suggest ensemble techniques are valuable for advancing drug discovery and development pipelines.