Astrogliosis has been recently investigated to identify genes that are over- or under-expressed and to derive the biological processes involved. In this line of research, we have considered a data set of gene expression in the neuroinflammation model induced by LPS and the ischemic stroke model (MCAO) (Zamanian et al., 2012), where standard biclustering methods fail to attribute a relevant role to the main metabolic pathways in separating models from controls. We analyze this data using DiVaMo, a method inspired by a covering-based feature selection approach (see Bertolazzi et al., 2016). The method identifies group of genes based on their integrated capability to differentiate between model and control samples, based on the expression ratios between models and controls. A single threshold allows a very straightforward control of the sensitivity and the robustness of the results. The separation power of a set is thus derived as a non-additive measure of the power of its genes; application to metabolism-related pathways (Glu/GABA, Glycolysis, TCA Cycle, PPP cycle, Lipid metabolism, NGF-TrkA/p75) identifies those that are significant for one of the two models, or for both.
Are metabolic processes affected during astrogliosis? DiVaMo, a non-standard analysis method, identifies new modifications in metabolic pathways in LPS and MCAO models of gliosis
G Felici;G Mavelli;P Bertolazzi
2016
Abstract
Astrogliosis has been recently investigated to identify genes that are over- or under-expressed and to derive the biological processes involved. In this line of research, we have considered a data set of gene expression in the neuroinflammation model induced by LPS and the ischemic stroke model (MCAO) (Zamanian et al., 2012), where standard biclustering methods fail to attribute a relevant role to the main metabolic pathways in separating models from controls. We analyze this data using DiVaMo, a method inspired by a covering-based feature selection approach (see Bertolazzi et al., 2016). The method identifies group of genes based on their integrated capability to differentiate between model and control samples, based on the expression ratios between models and controls. A single threshold allows a very straightforward control of the sensitivity and the robustness of the results. The separation power of a set is thus derived as a non-additive measure of the power of its genes; application to metabolism-related pathways (Glu/GABA, Glycolysis, TCA Cycle, PPP cycle, Lipid metabolism, NGF-TrkA/p75) identifies those that are significant for one of the two models, or for both.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
