Skip to main content
Article
The Effect of Heritability Estimates on High-Density Single Nucleotide Polymorphism Analyses with Related Animals
Journal of Animal Science
  • Abebe T. Hassen, Iowa State University
  • Santiago Avendano, Aviagen Ltd.
  • W. G. Hill, University of Edinburgh
  • Rohan L. Fernando, Iowa State University
  • Susan J. Lamont, Iowa State University
  • Jack C. Dekkers, Iowa State University
Document Type
Article
Publication Version
Published Version
Publication Date
1-1-2009
DOI
10.2527/jas.2008-0852
Abstract

Analysis of high-density SNP data in outbred populations to identify SNP that are associated with a quantitative trait requires efficient ways to handle large volumes of data and analyses. When using mixed animal models to account for polygenic effects and relationships, genetic parameters are not known with certainty, but must be chosen to ensure proper evaluation of SNP across chromosomes and lines or breeds. The objectives of this study were to evaluate the influence of heritability on the estimates and significance of SNP effects, to develop efficient computational strategies for analysis of high-density SNP data with uncertain heritability estimates, and to develop strategies to combine SNP test results across lines or breeds. Data included sire SNP genotypes and mean progeny performance from 2 commercial broiler breeding lines. Association analyses were done by fitting each SNP separately as a fixed effect in an animal model, using a range of heritabilities. The heritability used had a limited impact on SNP effect estimates, but affected the SE of estimates and levels of significance. The shape of the frequency distribution of P-values for the test of SNP effects changed from a highly skewed L-shaped curve at low heritability to a right-skewed distribution at high heritability. The P-values for alternative heritabilities could, however, be derived without reanalysis based on a strong linear relationship (R2 = 0.99) between differences in log-likelihood values of models with and without the SNP at different levels of heritabilities. With uncertain estimates of heritability, line-specific heritabilities that ensure proper evaluation of SNP effects across lines were determined by analysis of simulated sire genotypes and by permutation tests. Resulting heritability estimates were between those obtained from the entire breeding populations and those obtained from the data included in the sample data set. In conclusion, the uncertainty of heritability estimates has a limited impact on SNP effect estimates in association analyses, but a large impact on significance tests. The impact of heritability on tests can, however, be dealt with in a computationally efficient manner by using the strong linear relationship between model statistics under alternate levels of heritability. These approaches allow efficient analysis of large numbers of SNP for multiple traits and populations and pooling of results across populations.

Comments

This article is from Journal of Animal Science 87 (2009): 868, doi:10.2527/jas.2008-0852. Posted with permission.

Copyright Owner
American Society of Animal Science
Language
en
File Format
application/pdf
Citation Information
Abebe T. Hassen, Santiago Avendano, W. G. Hill, Rohan L. Fernando, et al.. "The Effect of Heritability Estimates on High-Density Single Nucleotide Polymorphism Analyses with Related Animals" Journal of Animal Science Vol. 87 Iss. 3 (2009) p. 868 - 875
Available at: http://works.bepress.com/susan_lamont/64/