Robust cluster analysis of microarray gene expression data with the number of clusters determined biologically

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Cluster analysis success relies on accurate expression pattern modeling.
• Robust methods are essential for handling violations of model assumptions in gene expression data.
• Outlier-resistant and distribution-insensitive clustering enhance reliability.

Purpose of the Study:

• To introduce a novel dissimilarity measure combining Euclidean distance and correlation coefficient advantages.
• To develop robust graphical and biological methods for summarizing and validating cluster analysis results.
• To compare the performance of rank-based versus log-based clustering approaches.

Main Methods:

• A new dissimilarity measure integrating Euclidean distance and correlation coefficient properties.
• Implementation of a rank order correlation coefficient for enhanced robustness.
• Development of a robust graphical summarization technique.
• Introduction of a biological method for determining the optimal number of clusters.

Main Results:

• The novel dissimilarity measure demonstrates improved robustness.
• Rank-based clustering methods exhibit superior performance compared to log-based methods in analyzing public gene expression data.
• The proposed graphical and biological methods effectively summarize and validate clustering outcomes.

Conclusions:

• Rank-based cluster analysis provides a more reliable approach for gene expression data.
• The developed methods offer robust tools for analyzing complex biological datasets.
• The findings suggest improved accuracy and interpretability in gene expression studies.