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

Is it better to combine predictions?

R D King1, M Ouali, A T Strong

  • 1Department of Computer Science, University of Wales, Aberystwyth Penglais, Aberystwyth, Ceredigion, SY23 3DB, Wales, UK.

Protein Engineering
|February 19, 2000
PubMed
Summary

Combining protein secondary structure prediction methods significantly improves accuracy by approximately 3%. Learning-based combination techniques, such as neural networks, outperformed simple voting methods on independent datasets.

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

Death of a female 'drug mule' from peritonitis: A case report and literature review.

The Medico-legal journal·2026
Same author

Xanthine oxidoreductase expression is diminished in breast cancer as a response to uric acid mediated chelation of redox active iron.

Free radical biology & medicine·2025
Same author

Cerebral small vessel disease burden and longitudinal cognitive decline from age 73 to 82: the Lothian Birth Cohort 1936.

Translational psychiatry·2021
Same author

Prioritisation of areas for early detection of southward movement of arctic fox rabies based on historical surveillance data in Quebec, Canada.

Epidemiology and infection·2020
Same author

Prevalence of antibiotic resistance and virulence genes in the biofilms from an aquifer recharged with stormwater.

Water research·2020
Same author

Scientific Business Abstracts of the 113th Annual Meeting of the Association of Physicians of Great Britain and Ireland.

QJM : monthly journal of the Association of Physicians·2019

Area of Science:

  • Computational biology
  • Bioinformatics
  • Structural bioinformatics

Background:

  • Accurate protein secondary structure prediction is crucial for understanding protein function and structure.
  • Individual prediction methods (PHD, DSC, NNSSP, Predator) show varying accuracies.

Purpose of the Study:

  • To compare the accuracy of individual protein secondary structure prediction methods.
  • To evaluate the effectiveness of combining predictions from multiple methods.
  • To identify optimal strategies for combining predictions.

Main Methods:

  • Comparison of four individual protein secondary structure prediction methods: PHD, DSC, NNSSP, and Predator.
  • Testing various combination strategies: voting, biased voting, linear discrimination, neural networks, and decision trees.

Related Experiment Videos

  • Training and validation using independent datasets, including CASP3 and EBI datasets.

    • Main Results:

    • NNSSP was the most accurate individual method, followed by PHD, DSC, and Predator, though differences were not statistically significant.
    • Combining predictions consistently improved accuracy by approximately 3% on both test datasets.
    • Learning-based combination methods (linear discrimination, neural networks) significantly outperformed voting techniques on the EBI dataset.

    Conclusions:

    • Combining predictions from multiple protein secondary structure prediction methods is superior to using individual methods.
    • Learning-based approaches offer significant advantages over simple voting for prediction combination.