Advantages and pitfalls in the application of mixed model association methods

By Jian Yang, Noah A. Zaitlen, Michael E. Goddard, Peter M. Visscher, Alkes L Price. Nature Genetics, 2014

967 citations recorded by the source corpus.

Mixed linear models are emerging as a method of choice for conducting genetic association studies in humans and other organisms. The advantages of the mixed-linear-model association (MLMA) method include the prevention of false positive associations due to population or relatedness structure and an increase in power obtained through the application of a correction that is specific to this structure. An underappreciated point is that MLMA can also increase power in studies without sample structure by implicitly conditioning on associated loci other than the candidate locus. Numerous variations on the standard MLMA approach have recently been published, with a focus on reducing computational cost. These advances provide researchers applying MLMA methods with many options to choose from, but we caution that MLMA methods are still subject to potential pitfalls. Here we describe and quantify the advantages and pitfalls of MLMA methods as a function of study design and provide recommendations for the application of these methods in practical settings.

Subfields: Genetics

View Paper: https://www.semanticscholar.org/paper/f8bd430f43c2c9e0c57afcd8d90832efeba62129

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Genetics

Citations in the corpus

Dissecting the genetics of complex traits using summary association statistics (2016, methodology): “The prediction R 2 for each method are given by 80,113 These equations can naturally be extended to linked markers (using the effective number of unlinked markers 104 ) and case-control traits (using observed-scale SNP-heritability 114 ).”
Enabling Privacy-Preserving GWAS in Heterogeneous Human Populations (2016, neutral): “Such analyses are complicated by systematic differences among different human populations (Yang et al., 2014).”
Enabling Privacy Preserving GWAS in Heterogeneous Human Populations (2016, methodology): “Population stratiﬁcation is the name given to systematic genomic diﬀerences between human populations [10].”
A generalized linear mixed model association tool for biobank-scale data (2021, methodology): “In our previous work, we developed an LMM-based GWA tool, fastGWA, that is orders of magnitude faster than other LMM-based tools, mainly owing to the use of a sparse genomic relationship matrix (GRM) to capture pedigree relatedness among individuals 10 .”
Iterative Usage of Fixed and Random Effect Models for Powerful and Efficient Genome-Wide Association Studies (2016, methodology): “Similar to previous comparisons[6,18,20], we decomposed the total computational time into three components: building the K, attaining optimization on variance components, and performing association tests.”
rMVP: A Memory-efficient, Visualization-enhanced, and Parallel-accelerated Tool for Genome-wide Association Study (2020, background): “To efficiently detect more candidate 36 genes with lower false positive rates is the current working goal for GWAS algorithms and 37 tools[2, 3].”
Genome-wide association studies in ancestrally diverse populations: opportunities, methods, pitfalls, and recommendations. (2019, background): “Modelling the genetic similarity between individuals can reduce false positives and improve power, with the increase in power proportional to the amount of variance explained by the modelled random effects terms (Yang et al., 2014).”
Mixed model association for biobank-scale data sets (2018, methodology): “Second, conditioning on genome-wide signal can produce loss of power under casecontrol ascertainment [2, 3]; specialized LMM methods are needed for modeling this scenario at scale.”
Computationally efficient whole-genome regression for quantitative and binary traits (2020, methodology): “This approach has been used to analyze COVID-19 outcomes across four studies and four ancestries 38 , and proved vital to provide accurate effect-size estimates for the meta-analysis.”
Using deep learning to model the hierarchical structure and function of a cell (2018, background): “State-of-the-art meth-ods in statistical genetics 32,33 are based on linear regression of phenotype against the independent effects of genetic polymorphisms, without modeling the underlying molecular mechanisms that give rise to nonlinearity and genetic interaction.”