Please activate JavaScript!
Please install Adobe Flash Player, click here for download


ISCB 2014 Vienna, Austria • Abstracts - Oral Presentations 77Wednesday, 27th August 2014 • 16:00-17:30 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust Wednesday, 27th August 2014 – 16:00-17:30 Invited session I6 Statistical methods for poly-omics studies Organizers: Axel Benner and Manuela Zucknick I6.1 From associations to mechanical understanding - data integration and causal inference in genomics R Spang1 1 University of Regensburg, Regensburg, Germany   If we want to find out whether a drug is effective in a certain disease, we have only one working option: We must test it in cellular assays, in mice and ultimately in clinical studies. Genomic reasoning is no option yet, but could it become one in the future? The problem requires a functional un- derstanding of cells and organisms. In more statistical terms, we need to infer causal relations between perturbations of cellular pathways and their downstream effects. In this talk I will give a brief introduction into cellular signaling and will then address a couple of statistical problems associated with their analy- sis: The detection of pathway activation in expression profiles of tumors, the construction of signaling models from perturbation data, and the esti- mation of causal effects from observational data.   I6.2 Bayesian models for integrative genomics M Vannucci1 1 Rice University, Houston, United States   Novel methodological questions are being generated in the biological sci- ences, requiring the integration of different concepts, methods, tools and data types. Bayesian methods that employ variable selection have been particularly successful for genomic applications, as they allow to handle situations where the amount of measured variables can be much greater than the number of observations. In this talk I will focus on models that integrate experimental data from different platforms together with prior knowledge. I will look in particular at hierarchical models that relate genotype data to mRNAs, for the selec- tion of the markers that affect the gene expression. Specific sequence/ structure information will be incorporated into the prior probability mod- els. All modeling settings employ variable selection techniques and prior constructions that cleverly incorporate biological knowledge about struc- tural dependencies among the variables. Applications will be to data from cancer studies.   I6.3 Do we gain by jointly analyzing multiple types of genomics data? WN van Wieringen1 1 VU University Medical Center, Dept. of Biostatistics, Amsterdam, The Netherlands   Through integration of genomics data from multiple sources, we obtain a more accurate and complete picture of the molecular mechanisms un- derlying tumorigenesis. Thus sounds the promise. What about practice? In this talk I will show that we may indeed gain from integrative analysis of the multiple genomics data. But adding clinical information to the mix proves valuable. To show joint analysis may deliver I concentrate on the integration of DNA copy number and gene expression data from oncogenomics studies with a two-sample set-up.These molecular levels are linked through the central dogma of molecular biology. In this context the aim is to identify differ- ential (between the two clinical groups) regulation among the genes of a pathway. For starters the gene-centered (univariate) analysis of such data is dis- cussed. This reveals no differential expression between the two groups. Alternatively, no significant association between the genomic and tran- scriptomic level is detected when ignoring group information. However, incorporate both clinical and genomic information and differen- tial associations abound. The main course features pathways. The interactions among the pathway´s molecular constituents are described by a structural equation model (SEM).With this model I am able to show that inclusion of DNA copy number data benefits the discovery of gene-gene interactions. Extension of the SEM to accommodate group information reveals differential regula- tion between the groups. But this differential gene-gene interaction pat- tern is missed when DNA copy number is not accounted for! Time for desert: is more thus better? Only when the data is well shaken and stirred.

Pages Overview