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

Recursive gene selection based on maximum margin criterion: a comparison with SVM-RFE.

Satoshi Niijima1, Satoru Kuhara

  • 1Department of Bioinformatics, Graduate School of Systems Life Sciences, Kyushu University, Hakozaki 6-10-1, Higashi-ku, Fukuoka 812-8581, Japan. niijima@grt.kyushu-u.ac.jp

BMC Bioinformatics
|December 26, 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

A multi-kingdom genetic barcoding system for precise clone isolation.

Nature biotechnology·2025
Same author

Identification and characterization of the suppressed lipid accumulation-related gene, SLA1, in the oleaginous yeast Lipomyces starkeyi.

Bioscience, biotechnology, and biochemistry·2024
Same author

INSTABILITY OF MESSENGER AND RIBOSOMAL RNA IN A GLUE-PROTEIN MUTANT OF Bombyx mori.

Development, growth & differentiation·2023
Same author

Development of a chub mackerel with less-aggressive fry stage by genome editing of arginine vasotocin receptor V1a2.

Scientific reports·2023
Same author

LsSpt23p is a regulator of triacylglycerol synthesis in the oleaginous yeast Lipomyces starkeyi.

Applied microbiology and biotechnology·2023
Same author

Prdm12 regulates inhibitory neuron differentiation in mouse embryonal carcinoma cells.

Cytotechnology·2022

Maximum Margin Criterion-Recursive Feature Elimination (MMC-RFE) offers improved gene selection for noisy microarray data. This method shows better performance for multi-class datasets compared to Support Vector Machine-Recursive Feature Elimination (SVM-RFE).

Area of Science:

  • Bioinformatics
  • Computational Biology
  • Genomics

Background:

  • Gene selection is crucial for improving prediction accuracy and identifying disease markers in microarray data analysis.
  • Support Vector Machine-Recursive Feature Elimination (SVM-RFE) is a popular gene selection method but is sensitive to noise and outliers in small sample datasets.
  • Microarray data often presents challenges due to noise and limited sample sizes, impacting the effectiveness of traditional gene selection techniques.

Purpose of the Study:

  • To propose a novel recursive gene selection method, Maximum Margin Criterion-Recursive Feature Elimination (MMC-RFE), to address limitations of existing methods.
  • To develop efficient and stable algorithms for MMC-RFE, extending naturally to multi-class prediction problems.
  • To compare the performance of MMC-RFE against SVM-RFE using various cancer microarray datasets.

Related Experiment Videos

Main Methods:

  • Developed Maximum Margin Criterion-Recursive Feature Elimination (MMC-RFE) utilizing the discriminant vector of the Maximum Margin Criterion (MMC).
  • Implemented efficient and stable algorithms to overcome computational drawbacks of classical Linear Discriminant Analysis (LDA) and handle high dimensionality.
  • Applied MMC-RFE and Support Vector Machine-Recursive Feature Elimination (SVM-RFE) to nine cancer microarray datasets, including four multi-class datasets.

Main Results:

  • MMC-RFE demonstrated competitive performance for binary-class datasets, falling between hard-margin and soft-margin SVM-RFE.
  • For multi-class datasets, MMC-RFE achieved significantly better performance than SVM-RFE using a smaller subset of genes.
  • The proposed MMC-RFE method showed reduced sensitivity to noise and outliers compared to SVM-RFE.

Conclusions:

  • MMC-RFE is a robust gene selection method, particularly effective for noisy and multi-class microarray data.
  • The method's resilience to noise suggests its utility in biomarker discovery from challenging datasets.
  • MMC-RFE offers a valuable alternative to SVM-RFE, especially when dealing with complex, real-world biological data.