Premium
Robust inference of genetic architecture in mapping studies
Author(s) -
Slate Jon
Publication year - 2017
Publication title -
molecular ecology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.619
H-Index - 225
eISSN - 1365-294X
pISSN - 0962-1083
DOI - 10.1111/mec.14052
Subject(s) - genetic architecture , biology , quantitative trait locus , evolutionary biology , inference , trait , association mapping , genetic association , selection (genetic algorithm) , computational biology , genome wide association study , genetic variation , genetics , variation (astronomy) , quantitative genetics , phenotype , linkage (software) , single nucleotide polymorphism , machine learning , artificial intelligence , gene , genotype , computer science , physics , astrophysics , programming language
The genetic architecture of a trait usually refers to the number and magnitude of loci that explain phenotypic variation. A description of genetic architecture can help us to understand how genetic variation is maintained, how traits have evolved and how phenotypes might respond to selection. However, linkage mapping and association studies can suffer from problems of bias, especially when conducted in natural populations where the opportunity to perform studies with very large sample sizes can be limited. In this issue of Molecular Ecology, Li and colleagues perform an association study of brain traits in ninespine sticklebacks Pungitius pungitius . They use a sophisticated approach that models all of the genotyped markers simultaneously; conventional approaches fit each marker individually. Although the single‐marker and multi‐marker approaches find similar regions of the genome that explain phenotypic variation, the overall conclusions about trait architecture are somewhat different, depending on the approach used. Single‐marker methods identify regions that explain quite large proportions of genetic variation, whereas the multi‐marker approach suggests the traits are far more polygenic. Simulations suggest the multi‐marker approach is robust. This study highlights how molecular quantitative genetics in wild populations can be used to address hypothesis‐driven questions, without making unrealistic assumptions about effect sizes of individual quantitative trait loci.