Welcome to the pages of the Center for Bioinformatics (Zentrum für Bioinformatik - ZBIT) Tübingen.

Takes place next Friday at 13h in A301, Sand 1. Have a look at the ZBIT colloquium talks archive.

Dr. Jörg Bernhardt, Institute for Microbiology, Ernst-Moritz-Arndt University Greifswald

Friday, 11.05.2012 at 13h (ct) in A301, Sand 1.  

Abstract: The field of data visualization in modern fields of omics is dominated by extensive lists of (more or less quantitative) data, mass spectra and little or more sophisticated Venn diagrams, bar and line charts. Besides such data displays also network representations (acyclic directed graphs) for illustrating distinguished interactions and tree like structures (tree graphs) for displaying gene/protein classifications are often applied.

Such tools are very powerful and suitable – for relatively small data sets of a few hundred bioentities. Their problem is: they need a lot of screen space making them difficult to handle for large data sets. Several thousand entities per data set need extensive scrolling or require navigation tools which again waste screen space. But the most valuable resource in data visualization is computer screen space. We have to answer the questions: How the immense data sets with genes / gene probes / proteins / peptides, uncertainty measures and expression levels can be displayed AND analyzed in an effective (space saving) but also intuitive way? How these data can be enriched with available biological knowledge.

Biological knowledge often is stored in systems of bio-ontologies e.g. open biological ontologies, Gene Ontology, TheSEED, KEGG Brite, Riley Schemes (TIGR classifications, GenoList classifications) and many more. The (quasi-)hierarchical structure of these classifications makes them suitable for tree based displays (wasting a lot of screen space in their traditional occurrence) OR even better for a treemap based display which are using any pixel of computer screen space for the simultaneous presentation of biological knowledge AND genes/proteins as well as probes/peptides, uncertainty measures and expression levels.

A Voronoi treemap based tool for the display of whole omics data sets from eukaryotes as well as from bacteria on one computer screen will be presented. We will illustrate how similarly regulated protein groups can be statistically secured and extracted by using Kolmogorov Smirnov statistics and how whole sets of expression profiles can be displayed by using genome and stream charts.