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

Selecting informative data for developing peptide-MHC binding predictors using a query by committee approach.

Jens Kaae Christensen1, Kasper Lamberth, Morten Nielsen

  • 1Center for Biological Sequence Analysis, BioCentrum-DTU, Technical University of Denmark, DK-2800 Lyngby, Denmark. jenskc@cbs.dtu.dk

Neural Computation
|November 25, 2003
PubMed
Summary

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

Review: application and opportunities for machine learning and artificial intelligence in preclinical immunogenicity risk assessment.

Frontiers in immunology·2026
Same author

Cancer epitope prediction tools and analysis pipelines in CEDAR.

Nucleic acids research·2026
Same author

NetMHCIIphosPan: A Machine Learning Tool for Predicting HLA Class II Antigen Presentation of Phosphorylated Peptides.

Journal of proteome research·2026
Same author

Standardising image registration and dose mapping for thoracic reirradiation: A national multi-centre benchmarking study.

Physics and imaging in radiation oncology·2026
Same author

NetMHCIIphosPan: a machine learning tool for predicting HLA class II antigen presentation of phosphorylated peptides.

bioRxiv : the preprint server for biology·2026
Same author

Integrated computational and experimental workflow identifies a novel T-cell epitope from Babesia bovis RON5 recognized by infected cattle.

Vaccine·2025

Query by Committee (QBC) effectively selects informative data points for training prediction algorithms, outperforming random selection and improving model accuracy for peptide-HLA-A2 binding affinity prediction.

Area of Science:

  • Computational biology
  • Machine learning
  • Immunoinformatics

Background:

  • Obtaining high-quality data for training prediction algorithms can be difficult and costly.
  • Effective data selection strategies are crucial for optimizing model performance.
  • Predicting peptide binding to MHC class I molecules, like HLA-A2, is vital in immunology.

Purpose of the Study:

  • To evaluate the effectiveness of a Query by Committee (QBC) strategy for selecting informative data points.
  • To compare QBC with random selection and high-affinity prediction strategies for training peptide-HLA-A2 binding affinity models.
  • To determine the optimal conditions for QBC strategy performance.

Main Methods:

  • Trained neural network algorithms to predict peptide binding affinity to HLA-A2.

Related Experiment Videos

  • Implemented and compared QBC strategy against random data selection and high-affinity prediction.
  • Analyzed the correlation between QBC values and measured binding affinities.
  • Investigated the impact of data quantity on QBC strategy effectiveness.

    • Main Results:

    • The QBC strategy significantly improved prediction performance compared to random selection.
    • Selecting peptides predicted to have high binding affinities also enhanced predictor accuracy.
    • QBC values correlated with measured binding affinities, indicating predictor disagreement on binding peptides.
    • QBC strategy performance is dependent on a sufficient amount of training data.

    Conclusions:

    • QBC is a superior strategy for selecting informative data points in costly data acquisition scenarios.
    • The QBC approach offers a data-efficient method for developing accurate peptide-HLA-A2 binding predictors.
    • The findings suggest broader applicability of QBC in other data-intensive scientific fields.