Parts | Days | Selection | Search | Downloads | Help
AKB: Biologische Physik
AKB 50: Poster Session "Biological Physics"
AKB 50.28: Poster
Friday, March 12, 2004, 10:30–13:00, B
Minkowski Functionals of high-dimensional data sets: application to protein analysis — •Boris Breidenbach1,2, Hubert Mantz1, and Klaus Mecke1,2 — 1MPI für Metallforschung, Heisenbergstr.3, 70569 Stuttgart — 2Universität Stuttgart, Institut für theoretische und angewandte Physik, Pfaffenwaldring 57, 70569 Stuttgart
The Minkowski functionals are a well established tool for the analysis of various types of spatial experimental data such as tomographic images of porous materials and atomic force measurements of thin polymer films where they have proven their ability to extract physically relevant information. Here, we present the first application of integral geometry in spaces with more than three dimensions: data is taken from microarray experiments of gene expression profiles in cells. Each protein is assigned a point in the high-dimensional concentration space. The functionals can be calculated exactly and efficiently for such decorated point distributions. The method is fast and statistically robust. We illustrate it by analyzing the gene-expression timecourse during the development of the Drosophila life cycle with respect to its clustering behaviour, i.e., its protein-protein interactions. The results are subsequently compared to standard techniques in bioinformatics such as hierarchical clustering and self-organising maps.