A0325
Title: Network inference and robust clustering on high-dimensional data to investigate molecular heterogeneity in glioma
Authors: Roberta Coletti - Center for Mathematics and Applications (NOVA Math), NOVA School of Science and Technology (Portugal) [presenting]
Marta Lopes - Department of Mathematics NOVA School of Science and Technology (Portugal)
Sofia Martins - NOVA School of Science and Technology - NOVA University of Lisbon (Portugal)
Abstract: Gliomas are a family of brain tumors that generally exhibit a low patient survival rate. Finding novel targets for personalized therapies necessitates a deeper knowledge of molecular underpinnings in various glioma types, which can be achieved by statistical and machine learning methods applied to the high-dimensional data sets nowadays generated. We propose a mathematical workflow to investigate differences and similarities in the three main glioma types: astrocytoma, oligodendroglioma, and glioblastoma. Based on gene expression data from The Cancer Genome Atlas, updated following the 2016 glioma classification guidelines, we estimated a sparse gene network for each glioma type by applying the joint graphical lasso algorithm. This led to a network-based variable selection, which was validated through robust sparse K-means clustering in different cases of study, to detect relevant genes in the separation of samples between classes in an unsupervised way. The outcomes disclose molecular differences between glioblastoma and the other glioma subtypes and point to potential novel glioma biomarkers needing further biological validation. The sets of selected variables appeared meaningful in the identification of robust clusters, though not totally in agreement with the preassigned diagnostic labels. This result supports further efforts towards the revision of the criteria for glioma classification.
