The Functional Genomics Facility and the Bioinformatics Research Centre at the University of Glasgow have recently introduced a set of new microarray analysis technologies. This website provides a brief summary of the essential components and links to references and software downloads. Rank ProductsThe first step in the interpretation of a microarray experiment is very often the detection of differentially expressed genes. Rank Products are a new test statistic that has been developed specifically for this purpose. They are particularly powerful for small and noisy datasets (with few replicates but many genes), which are typical of many microarray experiments. In that case they can often perform better than more traditional approaches (t-test, Wilcoxon rank sums, SAM).
Breitling R, Armengaud P, Amtmann A, Herzyk P (2004):
Rank products: A simple, yet powerful, new method to detect differentially regulated genes in replicated microarray experiments.
FEBS Letters 573: 83-92.
[PubMed]
[Supplement]
[Erratum]
[Download Perl implementation]
[Download Windows executable]
[Download BioConductor package]
[Faculty of 1000]
iterative GroupAnalysisThe second step in the analysis is then to make sense of the generated gene lists. This can be a tedious and also very subjective job, especially if the outcome of an experiment is unexpected. iGA is a technique that automates this step based on a combination of available gene annotation and the hypergeometric distribution. It identifies gene classes that are strikingly changed, and also highlights those class members that are particularly affected. Because class members can serve as “internal replicates” the iGA method can work even without experimental replication, and also with very noisy data. iGA assigns statistical confidence levels to its results and avoids many of the biases that are inherent in a manual analysis.
Breitling R, Amtmann A, Herzyk P (2004):
Iterative Group Analysis (iGA): A simple tool to enhance sensitivity
and facilitate interpretation of microarray experiments.
BMC Bioinformatics 5: 34.
[PubMed]
[Download Windows executable]
[Download Perl implementation]
[Manual]
[Addendum]
Please note that the filenames in the journal supplement are different from the filenames used in
the manual - rename the files accordingly for easier comparison with the instructions.
Graph-based iterative GroupAnalysisA particularly flexible and intuitive automated analysis of microarray results can be obtained by considering evidence networks. These are a graphical representation of knowledge about gene function, where each gene is connected to “related” genes based on available functional annotations. Graph-based iterative GroupAnalysis then identifies subgraphs that show particularly strong concerted changes and allows to displays them in an easily navigateable graphical format. This method can also be applied to metabolomics data, where it can be used to identified affected metabolic pathways based on metabolite screening results.
Breitling R, Amtmann A, Herzyk P (2004):
Graph-based iterative Group Analysis enhances microarray interpretation.
BMC Bioinformatics 5: 100.
[PubMed]
[Download Updated GiGA software]
FunAlyse PipelineMost of these techniques are now implemented in an automated analysis pipeline. Together with an RMA pre-processing step they are used as the standard analytical tool in the Sir Henry Wellcome Functional Genomics Facility. They are also available for outside users upon request.
Further ReadingThe first publication describing the application of many of these techniques to a new biological dataset is
Armengaud P, Breitling R, Amtmann A (2004):
The potassium-dependent transcriptome of Arabidopsis thaliana
reveals a prominent role of jasmonic acid in nutrient signaling.
Plant Physiol. 136: 2556-2576.
[PubMed]
[Supplementary Webpage]
We have also prepared a brief
summary of general microarray tips.
It is a subjective collection of issues identified during
a meta-analysis of data produced in the
SHWFGF.
PeopleRainer Breitling, now at Groningen Bioinformatics Centre, University of Groningen, The NetherlandsPawel Herzyk, Functional Genomics Facility and Bioinformatics Research Centre Anna Amtmann, Plant Science Group Patrick Armengaud, Plant Science Group Selected Users of GlaMA toolsAustralia Walter and Eliza Hall Institute of Medical Research Belgium Ghent University Hospital Brazil Instituto de Biologia Molecular do Parana Canada Biotechnology Research Institute Finland Turku Centre for Biotechnology Germany Charite, Humboldt-University BerlinUniversität Giessen Max Planck Institute for Developmental Biology Universität Tübingen Ireland University College Dublin Israel Agricultural Research Organization, Bet Dagan Italy University of Brescia Netherlands University of Groningen Wageningen University New Zealand University of Otago South Africa University of Pretoria Spain Instituto de Biotecnologia de Leon, INBIOTEC Instituto Valenciano de Investigaciones Agrarias, Moncada Switzerland University of Zurich Turkey Bogazici University, Istanbul United Kingdom University of Dundee University of Edinburgh, Medical School Health Protection Agency, Porton Down Imperial College London Institute of Food Research, Norwich University of Oxford Sainsbury Laboratory, John Innes Centre University of Wales, Swansea The Wellcome Trust Sanger Institute University of York Yoshitomi Research Institute of Neuroscience, Glasgow United States Creighton University Harvard Medical School Johns Hopkins Medical Institutions Oklahoma University Health Sciences Center Pennsylvania Hospital Pennsylvania State University The Research Institute for Children, Children's Hospital New Orleans The Salk Institute Stanford University St Jude Children's Research Hospital University of Arkansas for Medical Sciences University of California, San Diego University of California, San Franscico University of Chicago Hospitals University of Colorado at Boulder University of Utah
Last updated: 11/December/2005 |
[Rank Products are] a brilliant new approach to analysing microarray data.
Its main strength lies in the simplicity of the concept, which approaches the problem from a
biologist's point of view, rather than a statistician's. The outcome is a piece of software
which is particularly good at getting useful information even when the data are noisy.
A big advantage for those of us with limited resources is that reproducible results can be obtained with fewer
replicate experiments than is the case with other types of statistical treatment.
