Finding multiple coherent biclusters in microarray data using variable string length multiobjective genetic algorithm

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Microarray technology allows simultaneous monitoring of numerous gene expression patterns.
• Biclustering is crucial for identifying coregulated genes within specific experimental conditions.
• Effective biclustering requires identifying coherent, nontrivial biclusters with low mean squared residue and high row variance.

Purpose of the Study:

• To propose a multiobjective genetic biclustering technique for optimizing gene expression analysis.
• To develop a novel encoding scheme with variable chromosome length for improved biclustering.
• To introduce a new quantitative measure for evaluating bicluster quality.

Main Methods:

• A multiobjective genetic algorithm was employed to simultaneously optimize bicluster objectives.
• A novel variable-length chromosome encoding scheme was developed.
• A new quantitative measure was proposed to assess bicluster goodness.

Main Results:

• The proposed algorithm successfully identified coherent and nontrivial biclusters.
• Performance was validated on both simulated and real-life gene expression datasets.
• Comparison with existing biclustering techniques demonstrated competitive or superior results.

Conclusions:

• The developed multiobjective genetic biclustering technique offers an effective approach for gene expression data analysis.
• The novel encoding scheme and evaluation measure enhance the identification of coregulated gene groups.
• This method provides a valuable tool for uncovering complex regulatory patterns in genomics.