Feature Selection of OMIC Data by Ensemble Swarm Intelligence Based Approaches

Zhaomin Yao; Gancheng Zhu; Jingwei Too; Meiyu Duan; Zhiguo Wang

doi:10.3389/fgene.2021.793629

Feature Selection of OMIC Data by Ensemble Swarm Intelligence Based Approaches

Front Genet. 2022 Mar 8:12:793629. doi: 10.3389/fgene.2021.793629. eCollection 2021.

Authors

Zhaomin Yao^{1

2}, Gancheng Zhu³, Jingwei Too⁴, Meiyu Duan³, Zhiguo Wang^{1

2}

Affiliations

¹ Department of Nuclear Medicine, General Hospital of Northern Theater Command, Shenyang, China.
² College of Medicine and Biological Information Engineering, Northeastern University, Shenyang, China.
³ Key Laboratory of Symbolic Computation, College of Computer Science and Technology, Knowledge Engineering of Ministry of Education, Jilin University, Changchun, China.
⁴ Faculty of Electrical Engineering, Universiti Teknikal Malaysia Melaka, Hang Tuah Jaya, Melaka, Malaysia.

Abstract

OMIC datasets have high dimensions, and the connection among OMIC features is very complicated. It is difficult to establish linkages among these features and certain biological traits of significance. The proposed ensemble swarm intelligence-based approaches can identify key biomarkers and reduce feature dimension efficiently. It is an end-to-end method that only relies on the rules of the algorithm itself, without presets such as the number of filtering features. Additionally, this method achieves good classification accuracy without excessive consumption of computing resources.

Keywords: feature selection (FS); intersection and union combination; methylation data; swarm intelligence (SI); transcriptome data.