36 - Exploring the Statistical Nature of Independent Chromosome Segments

Tuesday, June 28, 2022

10:45 AM – 11:00 AM CT

Location: Oklahoma City Convention Center, 301 A/B

Presenting Author(s)

Mary Kate Hollifield, MS

Graduate Student
University of Georgia
Watkinsville, Georgia

Author(s)

Matias Bermann, n/a

P.h.D. student
University of Georgia
ATHENS, GA
Daniela Lourenco, PhD

Assistant Professor
University of Georgia
Athens, GA
Ignacy Misztal, n/a

Professor
University of Georgia
Athens, GA

Abstract: Independent chromosome segments for a population with N_e effective population size and L genome length can be approximately defined as 4N_eL non-overlapping haplotypes of L/2 length derived from any N_e animals. The number of independent chromosome segments (M_e) can be approximated as 4N_eL. The genetic selection with a genomic relationship matrix (GRM) using such haplotypes approaches that with a GRM using the SNP markers. The objective of this study was to investigate the statistical nature of independent chromosome segments. Data were simulated using QMSim and contained a population of ten non-overlapping generations, each including 2,000 animals with N_e equal to 20, and a polygenic trait with a heritability of 0.6. The last three generations were genotyped, and each genome contained ten 1 M long chromosomes, for a total genome length of 10 M and 50,000 SNP. Chromosome segments of each animal were organized in non-overlapping haplotypes by the SNP code using in-house software written in Fortran. The effects of the hypothetical independent chromosome segments were estimated using a model that included an overall mean plus the segment effects. To analyze the behavior around M_e, the number of segments chosen to estimate segment effects varied around 4N_eL. Accuracies were calculated for animals in the last generation by cor (TBV, Z_sŝ), where Z_sŝ is a vector of breeding values based on segments and TBV is a vector of the true breeding values outputted by QMSim. Accuracies of segment effects were compared with the true accuracy, i.e., cor(TBV,GEBV). The maximum accuracy of segment effects was 0.84, and the true accuracy was 0.96. The results suggest that 4N_eL segments contain most of the additive information in the population. However, the accuracy of GBLUP is greater than that of the chromosome segment effects, suggesting that arranging chromosome segments based solely on the statistical nature is not enough to account for all the genetic variation.