Event

---

---

---

09/11/2007 - 09/12/2007 | Günzburg

Statistics in Clinical Research: Challenges and Potentials

This tutorial provides an introduction to various applications of novel non-parametric methods based on u-statistics to the emerging field of molecular medicine. Each session will consist of (a) highlights of the method, (b) applications, and (c) exercises using R/S-Plus and Web-based bioinformatics tools accessible from http://muStat.rockefeller.edu. Participants are encouraged to bring their data sets to the course. In the linkage analysis part, introduction to the methodology will be followed by applications using GENEHUNTER or MERLIN.

Programme Directors:
Prof. Dr. E.M. Schneider, Director, Department of Experimental Anesthesiology, University of Ulm
Prof. K.M. Wittkowski, PhD, DSc, Head of Department for Experimental Design and Biostatistics, Center for Clinical and Translational Science, The Rockefeller University Hospital, New York, USA
PD Dr. J. Högel, Institute for Human Genetics, University of Ulm

Target group:
Students and medical doctors interested in biometrical analyses and involved in clinical studies and family segregation analysis.

Course Design:
Functional genetics and genomics have proven useful with rare diseases caused by single loci or single genes and resulting in binary phenotypes. Now, as the number of genetic, genomic, proteomic, and phenotypic variables increases, addressing more common diseases becomes feasible.
However, when a biological system is controlled by several co-regulated feedback loops, the functional form of the relationship between a measurement and genetic risk, disease activity, or treatment efficiency is typically unknown, so that the assumptions underlying statistical methods based on the linear model (either direct, as in variance, regression, discriminant, and factor analysis or indirect, as in neural networks) are typically not met.
This tutorial provides an introduction to various applications of novel non-parametric methods based on u-statistics to the emerging field of molecular medicine. Each session will consist of (a) highlights of the method, (b) applications, and (c) exercises using R/S-Plus and Web-based bioinformatics tools accessible from http://muStat.rockefeller.edu. Participants are encouraged to bring their data sets to the course. In the linkage analysis part, introduction to the methodology will be followed by applications using GENEHUNTER or MERLIN.

Teaching Methods: lecture, practical training

Objectives:
Non-parametric methods avoid artifacts due to unrealistic assumptions, yet rank-based methods can be computationally intensive (np-hard) when applied to multivariate data. Thus, traditional methods based on the linear model are still in use, despite their shortcomings. This tutorial presents computationally more efficient non-parametric methods based on u-statistics and bioinformatics tools available as packages for R/S-Plus (http://cran.r-project.org/src/contrib/Descriptions/muStat.html, http://csan.insightful.com/PackageDetails.aspx?Package=muStat) and MS Excel spreadsheets or Web servers (http://muStat.rockefeller.edu) to make these methods more widely available.
Further, participants should be able to understand the principles and main methods of genetic linkage analysis and to carry out standard linkage analyses using publicly available computer software.

Course Structure:
09.-11.09.2007 9:00-12:30, 13:30-17:00
Prof. Wittkowski: Nonparametric Statistics and Bioinformatics Tools for Genetics, Genomics, and Phenomics

12.09.2007 9:00-12:30; 13:30-17:00
PD Högel: Analysis of Human Genetic Linkage
Open end: After dinner opportunity will be given to further discuss and analyse individual data sets with the course directors

Course Contents:
Prof. Wittkowski: Nonparametric Statistics and Bioinformatics Tools for Genetics, Genomics, and Phenomics
o We will start by discussing variants of the oldest and most simple test statistic based on u-statistics, the McNemar or sign test. We will demonstrate why a family of frequently used analytic approaches for familial data in genetics, all based on the TDT, should be replaced by a stratified sign test, and how to further increase sensitivity through model selection. Practical exercises in R or S will be offered.
o As one of the first applications of Hoeffding's theoretical work on u-statistics, Gehan extended the Wilcoxon/Mann-Whitney ranksum test to censored data. We present new results on how to generalize this approach to more general multivariate data, including gene expression profiles and complex phenotypes. Exercises will demonstrate Excel spreadsheets that visualize the method using an approach developed 1914 by Deuchler in Tübingen, Germany.
o The first generation of statistical methods for analyzing microarray data is currently under review. We will present new tools, also based on u-statistics, for microarray quality control (detecting small artifacts on the surface of HDON microarrays) and for signal values estimation based on pairs of perfect and mis-matches (Affymetrix). These tools will be applied to different data sets using the R/Bioconductor package.
o We will extend this approach to a variety of partial orderings to allow for the various questions that may arise in genetics (diplotypes, epistasis), genomics (repeated measurements, such as changes in expression, depending on scale level), and phenomics (graded side effects, by attribution or severity). We will demonstrate access to R/S repositories (CRAN/CSAN) and a Web-server (http://muStat.rockefeller.edu) based on distributed processing for high-throughput screening.
PD Högel: Analysis of Human Genetic Linkage
o Aims and genetic fundaments of linkage analysis
o Parametric linkage analysis ("LOD-score")
o Nonparametric linkage analysis (IBD-sharing, "NPL-score")
o Practical computer training using test data sets and publicly available software

Information on participating / attending:
Course Fee:
Full participants: 500 € (including accommodation for 4 nights, 4 x breakfast, 4 x lunch, 3 x dinner, and coffee breaks); beverages have to be paid separately.
Students of the University of Ulm: 150 € (including 4 x lunch, 3x dinner, and coffee breaks, without staying overnight).
Students may apply for grants.

Deadlines:
Registration: 19.08.2007
Application for student's grants: 12.08.2007

Requirements:
Please bring your own laptop and data sets,
software to be used during the course should be downloaded and installed in advance:
GENEHUNTER-PLUS and ASM may be downloaded at http://galton.uchicago.edu/genehunterplus
MERLIN may be downloaded at http://www.sph.umich.edu/csg/abecasis/Merlin/download
A rich selection of further genetic analysis software can be found at http://linkage.rockefeller.edu/soft/list.html

Language: English/German according to audience's preference

Contact:
ICAS Secretariat
Prof. Dr. T.M. Fliedner, Dr. N. Selje-Aßmann, Ms. C. Steinwachs
Helmholtzstr. 20
D-89081 Ulm
Tel.: ++49-731-500-69405,
Fax: ++49-731-500-69406
e-mail: theodor.fliedner@uni-ulm.de
http://www.icas.uni-ulm.de

---