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

Avoiding model selection bias in small-sample genomic datasets.

Daniel Berrar1, Ian Bradbury, Werner Dubitzky

  • 1School of Biomedical Sciences, University of Ulster at Coleraine Northern Ireland. dp.berrar@ulster.ac.uk

Bioinformatics (Oxford, England)
|February 28, 2006
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

Genetic divergence of farmed Atlantic halibut (Hippoglossus hippoglossus) and potential for impact on wild populations.

BMC research notes·2026
Same author

Reference genome bias in light of species-specific chromosomal reorganization and translocations.

Genome biology·2025
Same author

Northern cod (Gadus morhua) movement: insights from acoustic telemetry and genomics.

Journal of fish biology·2025
Same author

A century of anthropogenic perturbations impact genomic signatures of the iconic migratory Atlantic cod.

Science advances·2025
Same author

Examining fish scale biomineral from Atlantic salmon populations.

Journal of fish biology·2024
Same author

Physiologic Determinants of Near-Infrared Spectroscopy-Derived Cerebral and Tissue Oxygen Saturation Measurements in Critically Ill Patients.

Critical care explorations·2024
Same journal

Biomedical Concept Recognition with Error-aware Negative-enhanced Ranking Framework.

Bioinformatics (Oxford, England)·2026
Same journal

TEDLH: Domain HMMs for sensitive detection of remote homologues.

Bioinformatics (Oxford, England)·2026
Same journal

PLNFGL: Joint Estimation of Multi-Condition Gene Networks from Single-cell RNA-seq Data.

Bioinformatics (Oxford, England)·2026
Same journal

MCFST: Spatial domain identification method based on multi-view graph convolutional network and graph fusion network.

Bioinformatics (Oxford, England)·2026
Same journal

SpaBiT: Enhancing Spatial Transcriptomics Resolution via Bidirectional Attention Transformers.

Bioinformatics (Oxford, England)·2026
Same journal

EDEL: Enhancing Dense Retrievers for Curation of Biomedical Knowledge Bases.

Bioinformatics (Oxford, England)·2026
See all related articles

Comparing classification models with small genomic datasets requires careful statistical methods. This study introduces a new approach to avoid optimistic bias in performance comparisons, ensuring reliable results for microarray data analysis.

Area of Science:

  • Bioinformatics
  • Computational Biology
  • Genomics

Background:

  • High-throughput technologies generate genomic datasets with many measurements per sample.
  • Classification models are frequently compared using resampling techniques for such data.
  • Existing comparison methods can lead to optimistically biased conclusions due to inadequate statistical control.

Purpose of the Study:

  • To address conceptual difficulties in comparative classification studies with genomic data.
  • To develop a statistically stringent framework for evaluating classifier performance.
  • To investigate bias in cross-validated model selection for small-sample scenarios.

Main Methods:

  • Comparison of various classifiers on multiclass microarray data.
  • Investigation of resampling techniques, including k-fold cross-validation and repeated random sampling.

Related Experiment Videos

  • Development of a novel statistical methodology to avoid bias in model selection.

    • Main Results:

    • Accuracy-based performance values are insufficient for comparing classifiers on small-sample genomic data.
    • The proposed statistical methodology effectively avoids bias in cross-validated model selection.
    • The methodology is applicable to both k-fold cross-validation and repeated random sampling.

    Conclusions:

    • Standard performance metrics are inadequate for comparing classifiers in small-sample genomic studies.
    • A statistically sound methodology is crucial for reliable classifier comparison.
    • The presented method offers a bias-free approach for model selection in cross-validation for genomic data analysis.