Background: Glioblastoma (GBM) is the most lethal primary brain cancer and its survival rate is very low. Comprehensive genomic characteristics and high degree of heterogeneity of GBM are main causes that contribute to the absence of effective therapeutic targets and prognostic markers and eventual treatment failures. Here, Gene set enrichment analysis (GSEA) was used to explore comprehensive genomic characteristics and high degree of heterogeneity of GBM. Our study will help explain potential tumorigenesis mechanism and contribute to development of targeted therapeutics and prediction of patient prognosis.Methods: Gene expression profile of GSE50161 was downloaded from Gene Expression Omnibus (GEO) database and GSEA was performed to evaluate the microarray data at level of gene sets of GO, KEGG and hallmark of Molecular Signatures Database (MSigDB). Core enrichments of selected hallmark gene sets were applied for clinical prognosis verification in Gene Expression Profiling Interactive Analysis (GEPIA). Genes associated with significant overall survival (OS) and unreported before were recognized as novel; the expression levels of novel genes in GBM samples and every novel gene between normal brain and GBM samples were compared. Receiver operating characteristic (ROC) and Pearson correlation analysis were also performed to evaluate the diagnostic biomarkers of GBM and correlations among novel genes respectively.Results: We obtained 511 and 494 GO gene sets, 12 and 10 KEGG gene sets, and 2 and 8 hallmark gene sets in normal and GBM samples respectively. Five novel genes SERPINA5, TENM2, ARAP3, THBD, TCF19 associated with GBM prognosis were selected and four novel genes SERPINA5, TENM2, ARAP3, TCF19 performed well for GBM diagnosis prediction were obtained. In addition, we also found that four novel genes SERPINA5, ARAP3, THBD, TCF19 that high expressed in GBM samples were all positively correlated with each other and correlation between ARAP3 and TCF19 was the strongest.Conclusions: Our study will help deeper understand the molecular heterogeneity and develop of targeted therapeutics of GBM; five novel related to prognosis genes (of which four are also related to diagnosis) may be applied to more accurate clinical decision-making process.