A comprehensive survey on computational learning methods for analysis of gene expression data

Area of Science:

• Genomics and computational biology
• Biomedical data science

Background:

• High-throughput gene expression analysis (microarray, RNA sequencing) generates vast datasets.
• Traditional statistical methods are insufficient for complex analyses like sample classification and feature gene discovery.
• Sophisticated computational approaches are essential for in-depth analysis of genomic and medical data.

Purpose of the Study:

• To compile and review statistical and computational tools for analyzing gene expression microarray data.
• To extend the applicability of discussed methods to RNA sequencing and quantitative proteomics datasets.
• To guide users in selecting appropriate data preprocessing and analysis methods based on desired outcomes.

Main Methods:

• Discussion of missing value imputation techniques.
• Overview of data normalization strategies.
• Detailed explanation of feature selection, feature extraction, and classification methods.
• Description of class discovery and evaluation parameters.

Main Results:

• Comprehensive compilation of diverse computational and statistical methods for gene expression data analysis.
• Adaptability of microarray analysis techniques to RNA sequencing and proteomics data highlighted.
• Structured approach to data preprocessing, analysis, and evaluation presented.

Conclusions:

• The review provides a detailed guide for selecting suitable computational and statistical methods for genomic and medical data analysis.
• Understanding various analytical tools empowers researchers to effectively process and interpret large-scale biological datasets.
• The presented methods facilitate advanced analyses such as sample classification and biomarker discovery.