Using Entropy as the Convergence Criteria of Ant Colony Optimization and the Application at Gene Chip Data Analysis

Area of Science:

• Computational Intelligence
• Bioinformatics
• Systems Biology

Background:

• Ant Colony Optimization (ACO) excels at shortest path problems but faces uncertainty in temporary solutions during iterations.
• Temporary solutions in ACO form a solution set, exhibiting entropy due to randomness.
• Solution set entropy stabilizes as the algorithm converges.

Purpose of the Study:

• To propose and validate entropy as a convergence criterion for Ant Colony Optimization (ACO).
• To approximate the optimal convergence time of ACO algorithms.
• To analyze gene expression data related to Alzheimer's Disease (AD) using the proposed method.

Main Methods:

• Introduced entropy as a convergence criterion for ACO.
• Applied the entropy-based ACO to cluster gene chip data from Alzheimer's Disease (AD) patients.
• Compared the performance of the entropy-converged ACO with six other clustering algorithms.

Main Results:

• The ACO algorithm utilizing entropy as a convergence criterion demonstrated high-quality clustering results.
• Clustering gene chip data from AD patients showed the superiority of the proposed method.
• Analysis of energy metabolism genes revealed a significant decrease in system entropy (disorder) with AD occurrence.

Conclusions:

• Entropy serves as an effective convergence criterion for ACO, approximating optimal convergence time.
• The entropy-based ACO algorithm is suitable for analyzing complex biological data, such as gene expression in AD.
• A decrease in the entropy of the energy metabolism system is a key characteristic observed during Alzheimer's Disease progression.