A8.21 Identification of geneexpression networks in different immunological states. (31st January 2014)

• Record Type: Journal Article
• Authors: Bonin, M; Kokatjuhha, J; Mans, K; Grützkau, A; Smiljanovic, B; Sörensen, T; Häupl, T
• Is Part Of: Annals of the rheumatic diseases. Volume 73:Supplement 1(2014)
• Page Start: A84
• Page End: A85
• Publication Date: 2014-01-31
• Subjects: Rheumatism -- Periodicals; 616.723005
• DOI: 10.1136/annrheumdis-2013-205124.195
• Languages: English
• ISSNs: 0003-4967

Abstract

Abstract : Background and Objective: Knowledge about gene networks is of great importance for analysis of transcriptome data. However, current tools mainly rely on information about direct molecular interactions between proteins, which is not directly connected to expression levels. These differences between transcriptome based perception of biological information and tools for network analysis are the main reason for difficulties in functional interpretation. Therefore, we started to use transcriptome data of biologically well-defined states to define functional markers and signatures as tools for future analysis. Materials and Methods: GeneChip HG-U133 Plus 2.0 transcriptomes from highly purified blood cell types (granulocytes, monocytes, CD4+ and CD8+ T-cell, B-cells, NK-cells) as well as from monocyte stimulation with LPS, TNF and type 1 IFN were selected from the BioRetis database (www.bioretis.de ). Correlations of expression between all probesets were calculated to filter for co-regulation Correlation matrices were calculated, clustered and displayed in heat maps The web-platform was constructed based on Ruby on Rails to provide a framework for analysis and storage of data The database and the correlation-algorithm will be provided on our homepage http://www.charite-bioinformatik.de . Results: Initially, correlation matrices were determined for each individual stimulation condition and its control. Stepwise combination of the three different conditions for calculation of correlation coefficients revealed a reduction of the correlation network and a reduction of overlap between the networks. This indicates increasing functional specificity of the identified candidates. All of the typical previously published IFN related genes were identified and thus confirmed our strategy. In a similar way, cell type specific co-expression networks were determined. Additional filtering for high signal intensity provides candidates for sensitive detection of the function related patterns even in highly diluted conditions. These marker panels are currently tested for detection and quantification of functional signatures in biopsies of inflamed tissue. Conclusions: Correlating transcription between genes in well-defined biological states identifies function-related markers and signatures. Depending on the type of function, appropriate conditions have to be selected.