Browsing by Subject "genomivalinta"

Sort by: Order: Results:

Now showing items 1-1 of 1
  • Martikainen, Katja (Helsingin yliopisto, 2015)
    In pig breeding many important traits are measurable only on females or after slaughter, which makes it difficult to observe these traits for the estimation of traditional estimated breeding values (EBVs). Genomic selection is expected to overcome these difficulties, since it makes use of large number of genetic markers called single nucleotide polymorphisms (SNPs). Dense genotyping of SNPs is needed for sufficient accuracy, but the costs increase with the density of the panel. It is possible to reduce the costs by genotyping some animals with a low-density SNP panels. Missing genotypes can then be imputed to correspond to the information of the denser SNP panel. The choice of optimal imputation program is dependent on structure of the population. The aim of this study was to compare performance of two imputation programs (BEAGLE and fastPHASE) to impute genotypes in the Finnish Yorkshire pig population. Data consisted of 809 boars. In data set 1 imputation was performed for all boars born after 2007 and in data set 2 for all boars born after 2005. Genotypes of the remaining boars were used as a reference population. Performance of these programs was evaluated by the allele error rate and computing time. The effect of SNP location, average distance between adjacent SNPs, size of the chromosome and MAF on the allele error rate was also studied. The average allele error rate using BEAGLE for the data set 1 was 2,88 % and time required for imputation was 8min 38s. Results using BEAGLE for the data set 2 were 2,58 % and 11min 50s. The average allele error rate using fastPHASE for the data set 1 was 4,02 % and time required for imputation for chromosome 1 was 1d 16h 11min 4s. Results using fastPHASE for the data set 2 were 3,71 % and 11d 11h 2min 31s. Allele error rates were highest at the end of the chromosome and lowest at the centre of the chromosomes. The average distance between adjacent SNPs did not have a notable effect on error rates. Error rates tended to be lower in large chromosomes than in small chromosomes. Error rates increased with increasing MAF. According to this study, BEAGLE is recommendable program for genotype imputation because of its good accuracy and short computational time.