Author:
Boulahfa Jawad,Floch Edith Le,Pierre-Jean Morgane,Deleuze Jean-François,Dandine-Roulland Claire
Abstract
Copy Number Variants (CNVs) are sometimes used to perform association studies. The aim of this paper was to study the use of CNVs in another context: heritability estimation. We wanted to assess the impact of using CNVs in these estimates, either alone, or in conjunction with Single Nucleotide Polymorphisms (SNPs). Using real SNP and CNV data from UK Biobank, we simulated phenotypes depending either on one or the two type(s) of data. We showed that mixed models, usually used for estimating heritability on SNP data, were also capable of estimating CNV heritability and to properly decipher between CNV and SNP heritabilities when phenotypes depend on both types of data. However CNV heritability estimation becomes more challenging when it is only supported by the few relatively common CNVs. Finally we estimated CNV and SNP heritabilities for two real phenotypes from UK Biobank (height and hypertension) but only hypertension showed a small but non-null CNV heritability of about 1.7%.
Publisher
Cold Spring Harbor Laboratory