JOINT ST ANDREWS/HIGHLANDS RSS GROUP MEETING
Wednesday 14 May
Lecture Theatre D,
Mathematical Institute,
North Haugh, St Andrews
(for updates please go to
http://www.mcs.st-and.ac.uk/StatsSeminars/index.shtml,
maps of the town and university can be found at
http://www.st-andrews.ac.uk/maps)
3:00: Professor Peter Bickel (University of California, Berkeley;
visiting the Newton Institute, Cambridge)
"Regularized covariance matrix estimation"
Abstract:
I shall review and discuss some of the different themes of
regularized estimation of the population covariance matrix:
1. Why estimate it and in what norm?
2. Pathologies of the empirical covariance matrix
3. Notions of sparsity and methods of regularization
4. Results of Bickel and Levina (2004-07)
5. Some future directions
4:00: Tea (in the Staff Room)
4:30: Professor Hans C. van Houwelingen (Leiden University Medical
Centre; visiting the Newton Institute, Cambridge)
"Dealing with heterogeneity in genetic studies: what can we
learn from meta-analysis? "
Abstract:
Heterogeneity between and within studies is becoming an
important issue in genome-wide association studies (Ioannidis et al.,
2007). Application of traditional random effects meta-analysis can
lead to a dramatic loss of power. However, the question arises
whether the established random effects analysis is appropriate for
genetic studies. Before considering this question, we first go back
to heterogeneity between and within linkage studies and meta-analysis
of linkage studies and see what lessons can be learned that carry
over to association studies.
Next, we will define potential null hypotheses for the
meta-analysis of association studies and explore the power and
usefulness of different statistical tests in a simulation study. The
general conclusion will be that the well-known random effects model
for the observed effects in different studies can be used, but that
the proper null hypothesis differs from clinical trial meta-analysis.
References:
Ioannidis, J..P., Patsopoulos, N.A., Evangelou, E. (2007):
Heterogeneity in meta-analyses of genome-wide association
investigations. PLoS ONE. 2007 Sep 5; 2(9): e841.
van Houwelingen, H.C., Arends, L.R., Stijnen, T. (2002): Tutorial in
Biostatistics, Advanced methods in meta-analysis: multivariate
approach and meta-regression. Statistics in Medicine, 21 (4), 589-624.
The meeting will be followed at 6.30 p.m. by a 2-course meal at the
Doll's House Restaurant in Church Square, St Andrews. The menu is
available at http://www.dolls-house.co.uk. (Click on "menus" and go
to "early evening menu".)
If you intend to come to the meal, please inform Peter Jupp,
preferably by email ([log in to unmask]), before 5.00 p.m. on Tuesday
13 May.
******************************************************************************
Professor P. E. Jupp
School of Mathematics and Statistics
University of St. Andrews
North Haugh, St. Andrews tel: (44) 1334 463704
Fife, KY16 9SS fax: (44) 1334 463748
Scotland e-mail: [log in to unmask]
url: http://www.mcs.st-andrews.ac.uk/~pej/
The University of St Andrews is a charity registered in Scotland : No SC013532
******************************************************************************
|