A Markov chain-based feature extraction method for classification and identification of cancerous DNA sequences

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Rising cancer incidence necessitates advanced analytical methods.
• Understanding the genetic basis of cancer requires analyzing DNA sequences.
• High computational costs hinder comprehensive gene analysis.

Purpose of the Study:

• To develop a feature dimension reduction technique for DNA sequences.
• To improve the efficiency and accuracy of cancer gene discrimination.
• To leverage the Markovian property of DNA sequences for computational analysis.

Main Methods:

• Analysis of cancer data using DNA sequences.
• Application of Markovian property for feature dimension reduction.
• Utilizing non-linear Support Vector Machine (SVM) with RBF and polynomial kernels for classification.

Main Results:

• The proposed method successfully reduced feature dimensions while preserving discriminatory properties.
• The non-linear SVM classifier effectively discriminated between cancerous and non-cancerous genes.
• Experimental results demonstrated low computational overhead and high accuracy via 10-fold cross-validation.

Conclusions:

• Markovian-based feature reduction combined with non-linear SVM offers an efficient and accurate approach for cancer gene discrimination.
• The method is validated on research case studies.
• This technique presents a promising solution for analyzing complex cancer genomics data.