Classifying six glioma subtypes from combined gene expression and CNVs data based on compressive sensing approach

It is realized that a combined analysis of different types of genomic measurements tends to give more reliable classification results. However, how to efficiently combine data with different resolutions is challenging. We propose a novel compressed sensing based approach for the combined analysis of gene expression and copy number variants data for the purpose of subtyping six types of Gliomas. Experiment results show that the proposed combined approach can substantially improve the classification accuracy compared to that of using either of individual data type. The proposed approach can be applicable to many other types of genomic data.