Predicting survival outcomes in ovarian cancer using gene expression data. (30th March 2019)

• Record Type: Journal Article
• Authors: Ahn, TaeJin; Kang, Nayeon; Kim, Yonggab; Kim, Se Ik; Song, Yong-Sang; Park, Taesung
• Is Part Of: International journal of data mining and bioinformatics. Volume 21:Number 4(2018)
• Page Start: 339
• Page End: 351
• Publication Date: 2019-03-30
• Subjects: RNA-seq -- ovarian cancer -- TCGA -- the cancer genome atlas -- survival analysis; Data mining -- Periodicals; Bioinformatics -- Periodicals; 006.312
• Languages: English
• ISSNs: 1748-5673

Abstract

About 70% of ovarian cancer types are High-Grade Serous Ovarian Cancer (HGSOC). Early stage HGSOC has a survival rate of more than 90%, but most diagnoses reoccur that the overall survival rate is only 35%. To detect early ovarian cancer, many studies have attempted to identify HGSOC-associated genes. In this study, we endeavoured to identify HGSOC related genes from RNA-seq data in The Cancer Genome Atlas (TCGA). We further suggest that stable extraction of genes could overcome difficulties regarding the reproducibility of existing RNA-seq data using a new gene selection strategy by Leave-One-Out Cross Validation (LOOCV). This strategy showed better performance than a previous method, when evaluating the same data set. Using this method, we could also infer biologic functions of selected genes, but instead, subsets of samples associated with different subsets of genes. These findings suggest that multiple signalling pathways contribute to ovarian cancer patient survival.