European Mathematical Genetics Meeting, Munich, Germany, 14^th-15^th May 2009

Source: Annals of Human Genetics, Volume 73, Number 6, November 2009 , pp. 658-669(12)

Buy & download fulltext article:

You have access to the full text article on a website external to ingentaconnect.

Please click here to view this article on Wiley Online Library.

You may be required to register and activate access on Wiley Online Library before you can obtain the full text. If you have any queries please visit Wiley Online Library

Abstract:

Testing for genetic association in the presence of linkage

Jeanine J. Houwing-Duistermaat, J. Lebrec and A. Callegaro

Dept of Medical Statistics and Bioinformatics, LUMC, Netherlands

j.j.houwing@lumc.nl

When dense single nucleotide polymorphism arrays have been typed in families, a strategy may be to first perform linkage analysis followed by association testing in regions showing linkage. We consider data on nuclear families, where parental genotypes are available. For this design, test statistics for association in the presence of linkage have been proposed for quantitative traits based on normally distributed random effects (Abecasis et al.) and for binary (Jonasdottir et al.) and survival (Zhong et al.) data using gamma distributed random effects. We derive test statistics for general phenotypes based on the generalized linear models framework where correlation between relatives is modeled using normally distributed random effects. We extend the methods developed by Lebrec et al. for linkage analysis of general phenotypes by adding a genotypic effect to the mean. For known identical by descent (IBD) status the variance of the test statistic is computed. When uncertainty in IBD exists we propose to estimate the variance empirically.

The new score statistic appears to be a weighted FBAT statistic. The performance of the new test statistic with respect to the standard FBAT statistic is studied by means of simulation. As illustration the test statistic is applied to data on Rheumatoid Arthritis from the NARAC study (GAW15, Witte et al.). From our simulations it appears that weighting by the IBD information only slightly improves the power. Improvement of power is obtained when the variance of the test-statistic is computed for situations where the IBD is known. References

Abecasis, G., Cardon, L. & Cookson,W. (2000) Am J of Hum Genet66, 279-292.

Jonasdottir, G., Humphreys, K. & Palmgren, J. (2007). Genet Epidemiol31, 528-540.

Lebrec, J. & Houwelingen, H. (2007) Human Heridity64, 5-15.

Witte, J.H., Schnell, A.H., Cordell, H.J., Almasy, L. & MacCCluer, J.W. (2007) Genet Epidemiol31, S1

Zhong, X. & Li, H. (2004). Biostatistics5, 307-327.

Document Type: Abstract

DOI: http://dx.doi.org/10.1111/j.1469-1809.2009.00547.x

Publication date: 2009-11-01