Protein-coding repeat polymorphisms strongly shape diverse human phenotypes

Abstract

Repeats associated with phenotype The degree to which repeated sequences within a genome affect human phenotypes has been difficult to establish. Mukamel et al . examined thousands of genomes in the UK Biobank and found that some of the largest effects of common genetic variants on human phenotypes, including those with clinical relevance, arise from protein-coding repeat polymorphisms (see the Perspective by Gymrek and Goren). Mapping the effects of the size and copy number of these repeated protein domains links genetic variation to human phenotypes, including lipoprotein(a) concentration, height, and male pattern balding. Furthermore, the alleles and frequencies of these repeated sequences differ between individuals of African and European descent, resulting in differences between the populations with clinical relevance for traits including lipoprotein(a) levels, a risk factor for coronary artery disease. —LMZ

Meta-analysis fine-mapping is often miscalibrated at single-variant resolution

“…These limitations are not specific to meta-analysis fine-mapping, and separate fine-mapping methods based on bespoke imputation references have been developed ( e . g ., HLA 81 , KIR 82 , and variable numbers of tandem repeats [VNTR] 83 ).…”

“…On the other hand, we find it challenging to use a LD reference when true causal variants are located within a complex region (e.g., major histocompatibility complex [MHC]), or are entirely missing from standard LD or imputation reference panels, especially for structural variants. These limitations are not specific to meta-analysis fine-mapping, and separate fine-mapping methods based on bespoke imputation references have been developed (e.g., HLA 81 , KIR 82 , and variable numbers of tandem repeats [VNTR] 83 ).…”

Phenotypic associations with the HMOX1 GT(n) promoter repeat in European populations

“…Recent studies have suggested that variable number tandem repeats have differing effects on differing haplotypic backgrounds, although this has not been shown for HMOX1. 42 Thirdly, although we tested our imputation approach extensively, and are confident of the accuracy to within 2 repeats, it may be that the imputation is biased with respect to important outcomes. This may be particularly relevant if for example, a set of SNPs that are causal for the outcomes and are usually coincident with an increased repeat length are systematically imputed incorrectly.…”

Similarity-Based Analysis of Allele Frequency Distribution among Multiple Populations Identifies Adaptive Genomic Structural Variants

“…Emerging technologies have recently revealed hundreds of thousands of genomic structural variants (SVs), including polymorphic duplications, deletions, inversions, and mobile transposable elements in the human genome ( Hurles et al 2008 ; Conrad et al 2010 ; Pang et al 2010 ; Mukamel et al 2021 ). Unlike single-nucleotide variants, each SV affects a continuous block in the genome and thus is more likely to result in a phenotypic effect ( Hurles et al 2008 ; Weischenfeldt et al 2013 ; Sudmant, Rausch, et al 2015 ).…”

“…Unlike single-nucleotide variants, each SV affects a continuous block in the genome and thus is more likely to result in a phenotypic effect ( Hurles et al 2008 ; Weischenfeldt et al 2013 ; Sudmant, Rausch, et al 2015 ). Several SVs have been documented to have considerable effects on human disease and evolution ( Dennis and Eichler 2016 ; Payer et al 2017 ; Hsieh et al 2019 ; Ho et al 2020 ; Mukamel et al 2021 ). Some of these functional variants reach >20% allele frequency in human populations, and some affect the copy number variation (CNV) of entire protein-coding genes ( McCarroll et al 2005 ; Handsaker et al 2015 ).…”