Bias in genomic predictions for populations under selection

• DOI: 10.1017/s001667231100022x
• Publication Date: 2011-07-18T15:29:17Z
• Authors: Z. G. VITEZICA; I. AGUILAR; I. MISZTAL; A. LEGARRA
• Keywords: Genomic Selection

Summary Prediction of genetic merit or disease risk using marker information is becoming a common practice for selection livestock and plant species. For the successful application genome-wide marker-assisted (GWMAS), genomic predictions should be accurate unbiased. The effect on bias accuracy was studied in two simulated animal populations under weak strong with several heritabilities. values by best-linear unbiased prediction (BLUP) data either from relatives summarized pseudodata genotyped individuals (multiple-step method) all available jointly (single-step method). single-step method combined genomic- pedigree-based relationship matrices. Predictions multiple-step were biased. less biased more but accurate. When relationships shifted constant, most value that which adjusts non-random individuals, can derived analytically.