Comprehensive vertical sample-based KNN/LSVM classification for gene expression analysis

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Microarray gene expression data analysis is crucial for identifying biological features and cell types in cancer diagnosis.
• High-dimensional gene expression data presents significant challenges for accurate and efficient classification.
• Existing methods struggle with the vast number of dimensions common in gene expression datasets.

Purpose of the Study:

• To develop a comprehensive classification approach for high-dimensional gene expression data.
• To improve both the accuracy and efficiency of classifying complex biological datasets.
• To provide a powerful tool for analyzing high-dimensional gene expression data in cancer research.

Main Methods:

• A vertical sample-based K-Nearest Neighbors (KNN) and Least Squares Support Vector Machine (LSVM) classification strategy was proposed.
• Genetic algorithms were employed to optimize the weights within the classification model.
• A novel data representation using P-trees and optimized logical algebra was utilized for enhanced speed.

Main Results:

• The proposed approach achieved high accuracy in classifying common gene expression datasets.
• Significant improvements in classification speed were observed due to the vertical data representation.
• The combination of KNN majority voting and local LSVM decision-making contributed to high accuracy.

Conclusions:

• The developed KNN/LSVM classification method offers a powerful solution for high-dimensional gene expression data analysis.
• The approach demonstrates the potential to enhance cancer diagnosis through accurate and efficient classification.
• The P-tree technology and optimized logical algebra significantly contribute to the method's efficiency.