Microarrays can provide a broad picture of the state of a cell by monitoring the expression level of thousands of genes at the same time. As such the technique can give valueable information on many biological processes. In particular in relation to tumors a precise classification is of importance for finding the optimal treatment.
Microarray data poses many interesting statistical problems. The raw data consists of an image of scanned pixel values and the location of a particular gene must be identified. Often there will be outliers in the data that need to be detected. A fundamental but very difficult question is the precise translation of a pixel value to the number of molecules present. Due to the variation in the production of the chips and the variation in the preparation of the samples the data from two arrays cannot be compared before a normalization has been made. This seems to be a non-trivial task and unless done very carefully one may introduce systematic errors. There seems to be a need of establishing an accepted standard for the normalization process. Coming to the estimation of the actual expression level of a gene one is still lacking a systematic study delineating the part of the variance due to the measurement technique and the part due to biological variation. Also for the Affymetrix system it is a very interesting question how to use best the information in the mismatch probes. When comparing two groups and searching for genes that are differentially expressed in the two groups one faces the problem of multiple comparisons. Obviously, if only a small number of genes participate in the difference between the two groups and we have no prior information as to what group of genes the differentially expressed genes belong to we are facing a problem when the chip contains a huge number of genes. Many microarray studies are used for classification and clustering. The maximum likelihood classifier based on independent normally distributed values seems to be one of the better choices. The assumptions for using this classifier needs to be checked and perhaps the modelling of the distributions can be improved. Underlying the above points are also the problem of designing the experiments so as best to extract the information.
It is the intention of the workshop to cover most of the aspects mentioned above.
|Wednesday February 19|
|19:00-21:00:||Registration and get-together|
|Thursday, February 20|
|09:40-10:25:||Claus Mayer (Rowett Research Institute, Aberdeen, Scotland): Least trimmed squares methods for microarray normalization|
|11:10-11:55:||David Edwards (NOVO Nordisk, Denmark): On the pre-analysis of one-channel cDNA microarray data|
|13:30-14:15:||Torben F. Ĝrntoft (Molecular Diagnostic Laboratory, Aarhus, Denmark): Microarrays, basic principle and use in medical research|
|14:30-15:15:||Yudi Pawitan (Department of Medical Epidemiology, Karolinska Institute, Stockholm): Survival analysis using gene expression data|
|16:00-16:45:||Mark van der Laan (University of California, Berkeley): Statistical inference with gene expression data|
|Friday, February 21|
|09:10-09:40:||Mette Langaas (Department of Mathematical Sciences, Norwegian University of Science and Technology, Trondheim, Norway): Estimating the number of genes truly differentially expressed|
|09:50-10:20:||Jörg Assmus (Department of Mathematics, University of Bergen, Norway): On the problem of significant p-values for gene expressions|
|11:00-11:45:||Jelle Goeman (Leiden University Medical Center, The Netherlands): A global score test for differential expression of groups of genes in high-dimensional microarray data|
|14:10-14:55:||Mats Rudemo (Mathematical Statistics, Chalmers University of Technology, Göteborg, Sweden): Empirical Bayes analysis of variance models for microarray data|
|15:40-16:25:||Ernst Wit (Department of Statistics, University of Glasgow, U.K.): Hidden Markov modelling of genomic expression interactions|
|16:40-17:25:||Arnoldo Frigessi (Norwegian Computing Center, Oslo, Norway): Towards a comprehensive statistical model of cDNA microarrays|
|Saturday, February 22|
|09:00-09:45:||Rafael Irizarry (Department of Biostatistics, Baltimore, USA): Getting usable data from microarrays: The role of statisticians|
|10:30-11:15:||Steen Knudsen (Technical University of Denmark): A new non-linear normalization method for reducing variablility in DNA microarray experiments|
|11:30-12:15:||Volkmar Liebscher (Institute of Biomathematics and Biometry, München, Germany): Stochasting modelling and quality control for gene expression data|
|13:15-14.00:||Anja von Heydebreck (Max-Planck-Institute for Molecular Genetics, Division of Computational Molecular Biology, Berlin, Germany): Error modelling, data transformation and robust calibration for microarray data|
The registration fee, covering lunches, is DKK 400, payable in cash in Danish currency upon arrival at the workshop.
Do not hesitate to contact the MaPhySto secretariat
for more information.
(This announcement in [ postscript-format | pdf-format ])