SGL-SVM: A novel method for tumor classification via support vector machine with sparse group Lasso.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Tumor classification is crucial in clinical medicine, with gene expression data presenting unique sparse characteristics.
• High-dimensional and small-sample tumor datasets pose challenges for accurate classification.

Purpose of the Study:

• To propose a novel tumor classification method, Sparse Group Lasso-Support Vector Machine (SGL-SVM), leveraging gene sparsity.
• To enhance the accuracy and efficiency of tumor classification using gene expression data.

Main Methods:

• Feature gene selection using Kruskal-Wallis rank sum test followed by Sparse Group Lasso (SGL).
• Classification performed using Support Vector Machine (SVM) on selected features.
• Validation on microarray and Next-Generation Sequencing (NGS) datasets, including BRCA and GBM.

Main Results:

• SGL-SVM demonstrated higher classification accuracy compared to other methods on multiple microarray and NGS datasets.
• The method effectively reduced the number of selected feature genes.
• Satisfactory performance was achieved through 10-fold and 5-fold cross-validation.

Conclusions:

• SGL-SVM is a promising method for classifying high-dimensional, small-sample tumor datasets.
• The approach effectively utilizes gene sparsity for improved tumor classification.
• The proposed method offers a valuable tool for clinical oncology research.