Gene selection for microarray cancer classification using a new evolutionary method employing artificial intelligence...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene selection is crucial for analyzing complex cancer microarray data.
• Identifying predictive genes remains a significant challenge in cancer classification.

Purpose of the Study:

• To propose a novel evolutionary method for identifying predictive genes for cancer classification.
• To enhance gene selection accuracy using genetic algorithms and artificial intelligence.

Main Methods:

• Applied a filter method for initial dimensionality reduction.
• Employed an integer-coded genetic algorithm with dynamic-length genotype and intelligent parameter settings.
• Investigated algorithmic behaviors, including convergence and parameter dynamics, and compared filter methods (Laplacian, Fisher score).

Main Results:

• The proposed evolutionary method was benchmarked on five high-dimensional cancer datasets.
• Statistical tests revealed significant differences in classifier and filter method performance across datasets.
• The method demonstrated superior performance compared to state-of-the-art approaches on the DLBCL dataset.

Conclusions:

• The novel evolutionary approach effectively identifies predictive genes for cancer classification.
• The method offers an advancement in handling complex, high-dimensional cancer microarray data.
• Further research can explore variations in filter methods and classifier choices for improved outcomes.