Luna Eris
Luna Eris
Luna Eris
Ivy Neurological Sciences Internship Program
School: Liberty High School
Hometown: Peoria, AZ
Mentor: Harshil Dhruv, PhD
Email This Article Print This Page

Refinement of Glioblastoma Subtypes by Contextual Gene Sets Identified by Context Mining Analysis

Glioblastoma Multiforme (GBM) is a highly invasive brain tumor. GBM does not always have a set Gene Expression Pattern (GEP), meaning few tumors are alike. Previous studies have discovered four subtypes of GBM based on GEP and clinical characteristics. However, the same standard cancer treatment exhibits wildly differentiated outcomes from patient to patient even within the same GBM subtype. Higher resolution definitions are desired in order to accurately diagnose and treat GBM. One way to further identify tumor subtypes is context mining. Context mining analysis identifies gene interaction networks from patient GEPs. If context mining analysis can identify detailed groups, and their signature genes are characterized, then a detailed GEP can be identified for many types of GBM. In this work we utilized the TCGA GBM data set to identify and characterize GBM molecular contexts. The GEPs of each identified GBM context were compared against all other GBM contexts to determine which genes are differentially expressed (DEG) for the selected GBM molecular contexts. From the 24 TCGA context groups, with a threshold of P <0.005, up to 200 of the most significant genes were selected as context signatures. Variation among the gene sets identified for each context highlights transcriptional heterogeneity of GBMs. Some context displays significantly smaller amount of DEGs, which may impact how they are defined in a clinical setting. In summary, we have identified a clear set of GEPs for each of the 24 context groups, which can be utilized to classify each GBM sample into higher resolution subgroups.