An improved algorithm for clustering gene expression data

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Microarray technology enables simultaneous monitoring of numerous gene expression levels over time.
• Analyzing this high-dimensional data requires robust clustering methods to address inherent uncertainty, noise, and imprecision.

Purpose of the Study:

• To propose a novel two-stage clustering algorithm for analyzing gene expression data.
• To address the challenge of data uncertainty and imprecision in microarray datasets.

Main Methods:

• A two-stage clustering approach combining a variable string length genetic scheme and a multiobjective genetic clustering algorithm.
• Incorporation of the concept of significant membership of points to multiple classes.
• Utilizing an iterated version of Fuzzy C-Means for clustering.

Main Results:

• The proposed two-stage algorithm demonstrated significant superiority over established methods like average linkage, Self-Organizing Map (SOM), and weighted Chinese Restaurant-based clustering (CRC).
• Validation was performed on both artificial and real-life gene expression datasets.
• The biological relevance of the obtained clustering solutions was assessed.

Conclusions:

• The developed two-stage clustering algorithm offers a more effective approach for gene expression data analysis.
• The method's ability to handle data uncertainty and its superior performance highlight its potential for biological discovery.