Miscellaneous

BioNet Reasoning

Command line application for modeling biological systems. It allows to reason, i.e. predict, explain, and plan, about biological systems that are described in the action language C_TAID.

general

Command line application for modeling biological systems. It allows to reason, i.e. predict, explain, and plan, about biological systems that are described in the action language C_TAID.

examples

Examples for knowledge bases, observations and queries describing the sulfur starvation response-pathway of Arabidopsis thaliana using the action language C_TAID are provided in examples.tar.gz.

dependencies

The Java application requires two additional software packages, lparse (lparse-1.0.17.tar.gz) and smodels (smodels-2.28.tar.gz). They are provided by the Helsinki University of Technology.

documentation

A comprehensive documentation of the action language C_TAID and the two supported translations can be found here (BioNetReasoningLong.pdf).
A short introduction to the syntax of the BioNetReasoning Tool and how it can be used to model biological systems can be found here (BioNetReasoningShort.pdf).

downloads

program file: BioNetReasoning.jar
examples: examples.tar.gz

CAPIU - Clustering using A Priori Information via Unsupervised decision trees

general

R library for CAPIU analysis for microarray experiments. CAPIU is a novel approach for clustering samples (treatments, patients, condition etc) by using annotational information about the genes. The algorithm searches all pre-defined gene classes for classes that exhibit a strong clustering of the samples. These are then used to split the samples in two groups until no significant splits can be found. The result is visualized as a tree with gene classes as nodes and groups of samples as leaves. For questions, comments, bugs etc, please contact Henning Redestig.

examples

Examples are provided in the R documentation files.

dependencies

R [tested with v2.2.0 on FC4 GNU/Linux]
Following packages are also needed: Biobase, MASS, mclust, e1071, cluster, hu6800, ellipse, GO.
dot, which is part of the Graphviz package.

documentation

The manuscript "Integrating functional knowledge during sample clustering for microarray data using unsupervised decision trees" has been accepted for publication in the Biometrical Journal. The Supplementary material is available here.

downloads
program file: capiu_0.2.tar.gz
R documentation: capiu.pdf

D2CMA

general

D2CMA contains a few R-scripts, mainly the function optimize.MA.design which builds upon the daMA package. It was used to optimize the two-color microarray experimental design for analyzing the differential responses of sensitive and tolerant rice cultivars to drought stress, as in Degenkolbe et al 2006 (Manuskript in prep., available upon request). For this kind of 2-color microarray experiment the optimal choice of sample pairs together with the labelling of the samples is dependent on the experimental question, i.e. the main effects of interest. For the example of Degenkolbe et al, we had two tolerant and two sensitive cultivars, hence our main interest was in main effect of drought stress (environmental effect) as well as environment by regulatory type (sensitive or tolerant). The coding of contrast matrices necessary to specify these effects of interest is as for the R package daMA. Hence, it is strongly recommended to become familiar with daMA prior to application of D2CMA functions. "optimize.MA.design" takes as input the description of overall design and the effects of interest, as well as possible hybridisations which are already performed, and optimizes the specified number of arrays left for hybridization with respect to minimal variance of the estimated effects (trace criterion, as recommended by the authors of daMA). The possibility of taking into account already performed hybridizations makes it a flexible tool to optimally react to changes during runtime of larger experiments.

For questions, comments, bugs etc, please contact Dirk Repsilber.

examples

Examples are provided in the R documentation files.

dependencies

R [tested with v2.1.1 and v2.3.1 on FC4 GNU/Linux]

performance

D2CMA function is *slow* -- optimizing a 2x4 design with 6 arrays already hybridized and 8 arrays left for optimization takes about 2h on an Intel® Xeon™ CPU 3.20GHz with 2 GB RAM

download D2CMA

 
loading content