Gene analysis for longitudinal family data using random-effects models

BMC Proceedings

Proceedings

Quinta Helmer¹  Bruna Balliu¹  Hae-Won Uh¹  Jeanine J Houwing-Duistermaat¹  Roula Tsonaka¹  Erik van den Akker² 

[1] Department of Medical Statistics and Bioinformatics, Leiden University Medical Center, PO Box 9600, 2300, Leiden, RC, The Netherlands;Department of Molecular Epidemiology, Leiden University Medical Center, PO Box 9600, 2300, Leiden, RC, The Netherlands;The Delft Bioinformatics Lab, Delft University of Technology, PO Box 5031, 2600, Delft, GA, The Netherlands;

关键词: Linear Mixed Model;    Rare Variant;    Functional Locus;    Gene Summary;    Variant Call Format;   

DOI  :  10.1186/1753-6561-8-S1-S88

来源: Springer

【 摘 要 】

We have extended our recently developed 2-step approach for gene-based analysis to the family design and to the analysis of rare variants. The goal of this approach is to study the joint effect of multiple single-nucleotide polymorphisms that belong to a gene. First, the information in a gene is summarized by 2 variables, namely the empirical Bayes estimate capturing common variation and the number of rare variants. By using random effects for the common variants, our approach acknowledges the within-gene correlations. In the second step, the 2 summaries were included as covariates in linear mixed models. To test the null hypothesis of no association, a multivariate Wald test was applied. We analyzed the simulated data sets to assess the performance of the method. Then we applied the method to the real data set and identified a significant association between FRMD4B and diastolic blood pressure (p-value = 8.3 × 10-12).

【 授权许可】

CC BY   
© Houwing-Duistermaat et al.; licensee BioMed Central Ltd. 2014