Rough Based Symmetrical Clustering for Gene Expression Profile Analysis.

Area of Science:

• Bioinformatics
• Computational Biology
• Data Mining

Background:

• Identifying coexpressed genes is crucial for understanding gene function in microarray data analysis.
• Point symmetry-based clustering is effective for recognizing complex cluster shapes in data.
• Existing methods face challenges with large datasets and determining the optimal number of clusters.

Purpose of the Study:

• To propose a distributed, time-efficient, and scalable parallel hybrid approach for point symmetry-based clustering of microarray data.
• To leverage rough set theory for faster convergence and automatic optimal classification.
• To address the challenge of an unknown number of clusters in gene expression data analysis.

Main Methods:

• A novel parallel hybrid approach combining point symmetry-based clustering with rough set theory and the K-means algorithm.
• Distributed and scalable implementation for handling large microarray datasets.
• Comparative analysis against other parallel symmetry-based K-means and existing K-means algorithms.

Main Results:

• The proposed algorithm demonstrates linear speedup in timing on large microarray datasets.
• Effective clustering of genes with similar symmetrical expression patterns.
• Statistical analysis confirms the significance of the new implementation, with biological relevance of clustering solutions also analyzed.

Conclusions:

• The developed parallel rough set-based hybrid approach offers an efficient and scalable solution for coexpressed gene identification in microarray data.
• This method overcomes limitations of existing algorithms by enabling automatic determination of cluster numbers and achieving significant speedups.
• The findings are validated through experiments on artificial, benchmark, and cancer gene expression datasets, highlighting its practical utility.