Stochastic Centre, Gothenburg arranges:

BIOINFORMATICS, GENETICAL STATISTICS, EPIDEMIC MODELLING

Miniworkshop, August 25

In connection with the large Functional Genomics Meeting in Gothenburg, August 26-27, we have arranged an afternoon, 1.30-5.00 pm, wednesday, August 25, with talks by some invited guests and local graduate students in mathematical statistics. The talks are open for everyone interested, (no fee or registration requested). It takes place at Matematiskt Centrum, Eklandagatan 86 (at the south-east corner of the Chalmers Campus) in lecture room MD 6.

Programme

Time Speaker Title
13.30-14.10 Arndt van Haeseler, Max-Planck-Institute for Evolutionary Anthropology, Leipzig, Germany Comparative Sequence Biology.
14.20-14.50 Tom Britton, Uppsala University Modelling and Analysis of Epidemics.
14.50-15.10Coffebreak
15.10-15.50 Bernard Prum, La Genopole, Evry, France Using Hidden Markov Models in the Analysis of Biological Sequences.
16.00-16.30 Mikael Knutsson, Chalmers Power Studies in Nonparametric Linkage Analysis, Combining Simulations and Multivariate Normal Approximations.
16.30-17.00 Staffan Nilsson, Chalmers Model Based Sampling and Weights in Affected Sib Pair Methods.

Abstracts of the talks

Arndt van Haeseler: Comparative Sequence Biology. (Joined work with Sonja Meyer and Gunter Weiss.)
Abstract: We have studied the complex pattern of nucleotide substitution in the control region of human mtDNA and of chimpanzee mtDNA hypervariable regions I. In order to elucidate the model of nucleotide substitutions that describes the evolution in an apropriate manner, a database was developed that allows an easy retrieval of the data.
We will briefly describe the set-up of the database and then explain the results obtained for humans and chimpanzees separately. Finally, the models of sequence evolution inferred for both species are compared. Thus, giving us insights, whether both species evolve according to the same model for the region studied.

Tom Britton: Modelling and Analysis of Epidemics.
Abstract: In the talk we will aim at giving a survey of the mathematical modelling, and its statistical analysis, for the spread of infectious diseases. In particular we will derive certain equations determining whether a major epidemic outbreak is possible or not, and in case of a long term outbreak what the endemic equlibrium will be. We will also discuss how model parameters may be estimated from data and how these estimates can be used in determining the necessary fraction to vaccinate in order to have the disease go extinct.

Bernard Prum: Using Hidden Markov Models in the Analysis of Biological Sequences.
Abstract: A way to look for information contained in DNA sequences (apart from the well known "genetic code") is to search "words" with a number of occurences higher (or smaller) than what can be predict.ed This can not be done without taking into account the number of sub-words of a word; this leads to work in Markov chain models (MC). Estimating the parameters of the MC shows differences between coding or non coding parts, as well as differences from one organism to another. Hence it is possible to suppose that a long sequence corresponds (in an unknow way) to successive (unknown) models. This is a Hidden Markov Model (HMM). Some example of the use of this tool will be shown, in particular concerning the search for horizontal transfers.

Mikael Knutsson: Power Studies in Nonparametric Linkage Analysis, Combining Simulations and Multivariate Normal Approximations.
Abstract: I will present a simulation method for power studies in linkage analysis using the NPL-score statistic calculated by the GENEHUNTER software. The method involves 3 steps; (1) Simulation of marker data given a specified genetic model, (2) NPL calculations using GENEHUNTER, and (3) Sampling from a multivariate normal distribution. Here, I will consider ASP families only and the method is illustrated and tested by a short example.

Staffan Nilsson: Model Based Sampling and Weights in Affected Sib Pair Methods.
Abstract: When running a genome scan with affected sib pairs and a nonparametric statistic in the form of sums of IBD counts, it turns out that, depending on the genetic model, some families are better than other if we consider the phenotypes of all family members. This can be utilized by performing selective sampling and/or putting weights on the IBD counts. Some different approaches are introduced and compared on simulated examples.

If you have questions, please contact nerman@math.chalmers.se

Welcome!

Olle Nerman, Mats Rudemo and Peter Jagers
Stochastic Centre, Gothenburg
Last modified: Tue Aug 31 10:30:22 MET DST 1999