Journal metrics

Journal metrics

  • IF value: 1.203 IF 1.203
  • IF 5-year value: 1.203 IF 5-year 1.203
  • CiteScore value: 0.98 CiteScore 0.98
  • SNIP value: 0.920 SNIP 0.920
  • SJR value: 0.390 SJR 0.390
  • IPP value: 0.89 IPP 0.89
  • h5-index value: 11 h5-index 11
  • Scimago H index value: 23 Scimago H index 23
Supported by
Logo Leibniz Institute for Farm Animal Biology Logo Leibniz Association
Arch. Anim. Breed., 61, 207-213, 2018
https://doi.org/10.5194/aab-61-207-2018
© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.
Original study
01 Jun 2018
The impact of QTL allele frequency distribution on the accuracy of genomic prediction
Pourya Davoudi1, Rostam Abdollahi-Arpanahi2, and Ardeshir Nejati-Javaremi1 1Department of Animal Science, University College of Agriculture and Natural Resources, University of Tehran, Karaj, 77871-31587, Iran
2Departments of Animal and Poultry Science, College of Aburaihan, University of Tehran, Pakdasht, 33916-53755, Iran
Abstract. The accuracy of genomic prediction of quantitative traits based on single nucleotide polymorphism (SNP) markers depends among other factors on the allele frequency distribution of quantitative trait loci (QTL). Therefore, the aim of this study was to investigate different QTL allele frequency distributions and their effect on the accuracy of genomic estimated breeding values (GEBVs) using best linear unbiased genomic prediction (GBLUP) in simulated data. A population of 1000 individuals composed of 500 males and 500 females as well as a genome of 1000 cM consisting of 10 chromosomes and with a mutation rate of 2.5  ×  10−5 per locus was simulated. QTL frequencies were derived from five distributions of allele frequency including constant, uniform, U-shaped, L-shaped and minor allele frequency (MAF) less than 0.01 (lowMAF). QTL effects were generated from a standard normal distribution. The number of QTL was assumed to be 500, and the simulation was done in 10 replications. The genomic prediction accuracy in the first-validation generation in constant, and the uniform allele frequency distribution was 0.59 and 0.57, respectively. Results showed that the highest accuracy of GEBVs was obtained with constant and uniform distributions followed by L-shaped, U-shaped and lowMAF QTL allele frequency distribution. The regression of true breeding values on predicted breeding values in the first-validation generation was 0.94, 0.92, 0.88, 0.85 and 0.75 for constant, uniform, L-shaped, U-shaped and lowMAF distributions, respectively. Depite different values of regression coefficients, in all scenarios GEBVs are biased downward. Overall, results showed that when QTL had a lower MAF relative to SNP markers, a low linkage disequilibrium (LD) was observed, which had a negative effect on the accuracy of GEBVs. Hence, the effect of the QTL allele frequency distribution on prediction accuracy can be alleviated through using a genomic relationship weighted by MAF or an LD-adjusted relationship matrix.
Citation: Davoudi, P., Abdollahi-Arpanahi, R., and Nejati-Javaremi, A.: The impact of QTL allele frequency distribution on the accuracy of genomic prediction, Arch. Anim. Breed., 61, 207-213, https://doi.org/10.5194/aab-61-207-2018, 2018.
Download
Share