Comparison of gene-based rare variant association mapping methods for quantitative traits in a bovine population with complex familial relationships

Genetics Selection Evolution

Research Article

Mario P. L. Calus¹  Bernt Guldbrandtsen²  Mogens Sandø Lund²  Goutam Sahana²  Qianqian Zhang³ 

[1] Animal Breeding and Genomics Centre, Wageningen UR Livestock Research, Wageningen, The Netherlands;Department of Molecular Biology and Genetics, Center for Quantitative Genetics and Genomics, Aarhus University, 8830, Tjele, Denmark;Department of Molecular Biology and Genetics, Center for Quantitative Genetics and Genomics, Aarhus University, 8830, Tjele, Denmark;Animal Breeding and Genomics Centre, Wageningen UR Livestock Research, Wageningen, The Netherlands;

关键词: Minor Allele Frequency;    Rare Variant;    Imputation Accuracy;    Polygenic Effect;    Sequence Kernel Association Test;   

DOI  :  10.1186/s12711-016-0238-5

 received in 2016-01-08, accepted in 2016-08-04,  发布年份 2016

来源: Springer

【 摘 要 】

Background There is growing interest in the role of rare variants in the variation of complex traits due to increasing evidence that rare variants are associated with quantitative traits. However, association methods that are commonly used for mapping common variants are not effective to map rare variants. Besides, livestock populations have large half-sib families and the occurrence of rare variants may be confounded with family structure, which makes it difficult to disentangle their effects from family mean effects. We compared the power of methods that are commonly applied in human genetics to map rare variants in cattle using whole-genome sequence data and simulated phenotypes. We also studied the power of mapping rare variants using linear mixed models (LMM), which are the method of choice to account for both family relationships and population structure in cattle.ResultsWe observed that the power of the LMM approach was low for mapping a rare variant (defined as those that have frequencies lower than 0.01) with a moderate effect (5 to 8 % of phenotypic variance explained by multiple rare variants that vary from 5 to 21 in number) contributing to a QTL with a sample size of 1000. In contrast, across the scenarios studied, statistical methods that are specialized for mapping rare variants increased power regardless of whether multiple rare variants or a single rare variant underlie a QTL. Different methods for combining rare variants in the test single nucleotide polymorphism set resulted in similar power irrespective of the proportion of total genetic variance explained by the QTL. However, when the QTL variance is very small (only 0.1 % of the total genetic variance), these specialized methods for mapping rare variants and LMM generally had no power to map the variants within a gene with sample sizes of 1000 or 5000.Conclusions We observed that the methods that combine multiple rare variants within a gene into a meta-variant generally had greater power to map rare variants compared to LMM. Therefore, it is recommended to use rare variant association mapping methods to map rare genetic variants that affect quantitative traits in livestock, such as bovine populations.

【 授权许可】

CC BY   
© The Author(s) 2016

【 图 表 】

12951_2015_155_Article_IEq26.gif

12951_2015_155_Article_IEq25.gif

12951_2015_155_Article_IEq24.gif

Fig. 2

12951_2017_270_Article_IEq5.gif

Fig. 3

Fig. 1

12951_2015_111_Article_IEq1.gif

Fig. 6

Fig. 4

Fig. 1

12951_2017_315_Article_IEq1.gif

Fig. 1

12951_2015_155_Article_IEq23.gif

12951_2015_155_Article_IEq22.gif

12951_2015_155_Article_IEq21.gif

12951_2015_155_Article_IEq20.gif

12951_2015_155_Article_IEq12.gif

Fig. 2

12951_2015_155_Article_IEq9.gif