3.1. Development and distribution of data analysis software.

Responsible: Tim Beissbarth, DKFZ, Heidelberg.

Background:

We began the development and distribution of data analysis software within the context of NGFN-1, with the goal of providing practical exercises for our Microarray Data Analysis courses and to furnish other researchers with the results of our theoretical work. Within the collaborative Bioconductor (www.bioconductor.org) framework we have produced and are maintaining numerous software packages. More specifically, we have developed a workflow oriented data-mining platform that integrates methods from statistics and machine-learning into one common framework. The general goal of this particular sub-project is thus to utilize this framework in enabling the experimental labs in the NGFN to analyse their profiling data themselves.

Planned work:

The primary goal of our work is to provide technologies that permit high-quality and standardised analysis of NGFN-2 data. Analysis questions are often complex and specialized towards particular biological or clinical applications and related methods are often equally complex and numerous. We therefore aim to provide standardized, re-usable workflows integrated into Mine-IT for distribution into NGFN nodes. To address the need for statistical anlysis of graph-like data sets (e.g. protein interaction data from protein arrays or mass spectrometry) and network-like metadata (e.g. regulatory or signalling pathways), we will further develop software modules for the statistical and graph theoretical analysis and visualisation of graphs. In particular, we will enhance the Bioconductor packages graph and Rgraphviz.