Abstract
The aim of this study was to investigate imputation quality from two widely used imputation software packages (Flmpute and findhap) in imputing from 50k to HD. Three different aspects with respect to imputation quality were investigated: (1) differences between software programs, (2) differences in the number of HD-genotyped ancestors for four different groups of animals and (3) differences in marker coverage between chromosomes. For this purpose the Illumina chip annotations of two 50k chips (BovineSNP5O vi and v2) and the HD chip (BovineHD) were examined to find large physical gaps between adjacent SNPs or areas on the chromosome with a low marker density. After that, both programs were tested under routine conditions to assess the completeness of imputation and the amount of computational resources required. It turned out that imputation quality is influenced by differences between chromosomes and by differences between groups of animals with respect to the degree of HD genotyped ancestors in the reference population. Both differences had a greater impact on the performance of flndhap as compared to Flmpute. Furthermore using Flmpute resulted in higher imputation accuracy in the considered categories of this study.
Translated title of the contribution | Imputation with Flmpute and findhap |
---|---|
Original language | German |
Pages (from-to) | 81-94 |
Number of pages | 14 |
Journal | Zuchtungskunde |
Volume | 86 |
Issue number | 2 |
State | Published - 2014 |
Externally published | Yes |