Hybrid Method Based on Information Gain and Support Vector Machine for Gene Selection in Cancer Classification

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene expression data presents challenges for cancer classification due to high dimensionality, small sample sizes, and noise.
• Accurate cancer classification is crucial for effective diagnosis and treatment strategies.

Purpose of the Study:

• To develop a hybrid gene selection method, Information Gain-Support Vector Machine (IG-SVM), to enhance cancer classification accuracy.
• To effectively filter irrelevant and redundant genes while minimizing noise in gene expression datasets.

Main Methods:

• A hybrid approach combining Information Gain (IG) for initial gene filtering and Support Vector Machine (SVM) for further redundancy removal.
• Utilizing the selected informative genes as input for the LIBSVM classifier.
• Evaluating the IG-SVM method on five diverse cancer gene expression datasets.

Main Results:

• The IG-SVM method demonstrated superior performance and higher classification accuracy compared to other algorithms.
• Achieved a 90.32% classification accuracy for colon cancer using only three genes (CSRP1, MYL9, GUCA2B).
• Successfully identified a minimal set of informative genes for accurate cancer classification.

Conclusions:

• The IG-SVM hybrid gene selection method is effective for improving cancer classification accuracy from high-dimensional gene expression data.
• This approach offers a robust and efficient way to identify key genes for cancer diagnosis.
• The findings suggest potential for simplified and more accurate cancer diagnostic tools.