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

Multi-class tumor classification by discriminant partial least squares using microarray gene expression data and

Yongxi Tan1, Leming Shi, Weida Tong

  • 1Burns and Allen Research Institute Microarray Core, Cedars-Sinai Medical Center, David Geffen School of Medicine, UCLA, Los Angeles, CA 90048, USA.

Computational Biology and Chemistry
|July 21, 2004
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

Distinct neural processes link speech planning and execution.

bioRxiv : the preprint server for biology·2026
Same author

Prenatal e-cigarette exposure induces region- and cell type-specific metabolic and developmental reprogramming.

bioRxiv : the preprint server for biology·2026
Same author

Benchmarking Q40 sequencing for sensitive and efficient detection of rare genomic variants.

Genome biology·2026
Same author

Astrocytic D1 Dopamine-Signaling Regulates Synaptic Remodeling and Cocaine Seeking.

bioRxiv : the preprint server for biology·2026
Same author

Palliative Care or Hospice? Flipping the Classroom for 1st Year Pre-clinical Medical Students With Interactive Online Content.

The American journal of hospice & palliative care·2026
Same author

Prenatal THC exposure disrupts mitochondrial respiratory gene programs and medium spiny neuron maturation trajectories in the nucleus accumbens.

bioRxiv : the preprint server for biology·2026

Discriminant partial least squares effectively classifies human tumors using gene expression data. Leave-half-out cross-validation offers a realistic estimate of predictive ability for molecular diagnostics.

Area of Science:

  • Genomics
  • Bioinformatics
  • Cancer Diagnostics

Background:

  • High-throughput DNA microarrays enable simultaneous monitoring of thousands of gene expression levels.
  • Microarray technology holds promise for molecular diagnostics, including cancer classification.
  • Analyzing high-dimensional gene expression data (thousands of genes, fewer samples) presents a significant challenge.

Purpose of the Study:

  • To develop and evaluate reliable classification methods for high-dimensional microarray data.
  • To accurately assess the predictive ability and reliability of classification models for cancer diagnosis.
  • To compare the performance of different cross-validation procedures in evaluating classification models.

Main Methods:

  • Utilized discriminant partial least squares (DPLS) for tumor classification.

Related Experiment Videos

  • Applied DPLS to four distinct human tumor microarray datasets.
  • Employed four cross-validation procedures: leave-one-out, leave-half-out, incomplete, and full.

    • Main Results:

    • Discriminant partial least squares demonstrated good prediction performance in classifying human tumors.
    • Leave-half-out cross-validation provided a more realistic estimate of predictive ability compared to other methods.
    • Certain cross-validation procedures may overestimate a classification model's predictive power.

    Conclusions:

    • Discriminant partial least squares is a viable method for molecular cancer diagnostics using gene expression data.
    • Leave-half-out cross-validation is recommended for a more accurate assessment of model performance.
    • Integrating information from multiple cross-validation techniques enhances the evaluation of classification model reliability.