Cancer subtype prediction from a pathway-level perspective by using a support vector machine based on integrated gene expression and protein network. (April 2017)

• Record Type: Journal Article
• Authors: Hung, Fei-Hung; Chiu, Hung-Wen
• Is Part Of: Computer methods and programs in biomedicine. Volume 141(2017)
• Page Start: 27
• Page End: 34
• Publication Date: 2017-04
• Subjects: Cancer subtype -- Protein–protein interaction -- Gene expression -- Signaling pathway -- Neuroepithelial tumor -- Computational method; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2017.01.006
• Languages: English
• ISSNs: 0169-2607

Abstract

Highlights: Genes and PPIs were integrated for the intensity of the pathway link change in the neuroepithelial tumors progress. Fragments of pathways in different stages of disease were transformed into a sparse matrix. The SVM prediction accuracy was 67.64% for 3 subtypes in neuroepithelial tumors. Lesser features than those applied by gene expression methods were used to obtain similar results. Abstract: Background and objective: Distinguishing cancer subtypes is critical for selecting the appropriate treatment strategy. Bioinformatics approaches have gradually taken the place of clinical observations and pathological experiments. However, these approaches are typically only used in gene expression profiling. Previous studies have primarily focused on the gene level or specific diseases, and thus pathway-level factors have not been considered. Therefore, a computational method that integrates gene expression and pathway is necessary. Methods: This study presented an approach to determine potential fragments of activated pathways around protein networks in different stages of disease. We used a scored equation that integrates genomic and proteomic information and determined the intensity of the pathway link change. A support vector machine (SVM) was used to train and test subtype-predicted models. Results: The performance of the proposed method was evaluated by calculating prediction accuracy. The average prediction accuracy was 67.64% for three subtypes in tumors of neuroepithelial tissues. The results demonstrate that the proposed method applies fewer features than gene expression methods used to obtain similar results Conclusions: This study suggests a method to implement a cancer subtype classifier based on an SVM from a pathway-level perspective.