Clustering algorithms: on learning, validation, performance, and applications to genomics.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Microarray technology allows simultaneous measurement of thousands of gene expressions.
• Data clustering, widely used in pattern recognition, is now a popular technique for analyzing microarray data.
• Clustering genes reveals co-expressed genes, while clustering samples aids in disease pathology discrimination.

Purpose of the Study:

• To provide a comprehensive overview of clustering techniques for microarray data analysis.
• To discuss theoretical aspects, popular algorithms, and validation indices for clustering.
• To evaluate the performance of different clustering approaches in computational biology.

Main Methods:

• Review of theoretical aspects of clustering, including error and learning.
• Overview of popular clustering algorithms (e.g., k-means, hierarchical clustering).
• Discussion of classical validation indices for assessing cluster quality.

Main Results:

• Clustering genes groups genes with similar expression profiles across samples.
• Clustering samples can identify distinct pathological groups based on gene expression patterns.
• The choice of algorithm and validation index significantly impacts results and requires careful consideration.

Conclusions:

• Clustering is a powerful tool for analyzing high-throughput microarray data in biology and medicine.
• Selecting appropriate clustering methods is critical due to the complexity and high dimensionality of biological data.
• This paper serves as a guide to understanding and applying clustering in computational biology.