26
Genome-wide association and genomic prediction of response to infection for two isolates of porcine reproductive and respiratory syndrome virus

Tuesday, March 17, 2015: 3:15 PM
302-303 (Community Choice Credit Union Convention Center)
Emily H. Waide , Iowa State University, Ames, IA
Nick V.L. Serão , Iowa State University, Ames, IA
Andrew Hess , Iowa State University, Ames, IA
Raymond R.R. Rowland , Kansas State University, Manhattan, KS
Joan K. Lunney , USDA, ARS, BARC, APDL, Beltsville, MD
Graham S. Plastow , University of Alberta, Edmonton, AB, Canada
Jack C. M. Dekkers , Iowa State University, Ames, IA
Abstract Text: The Porcine Reproductive and Respiratory Syndrome (PRRS) Host Genetics Consortium and Genome Canada projects aim to identify genetic loci associated with response to PRRS virus (PRRSV) infection. The objective of this study was to analyze data from 13 trials of ~200 nursery piglets infected with one of two PRRSV isolates, NVSL97-7985 (NVSL) and KS2006-72109 (KS06). Phenotypes included weight gain (WG) from infection to 42 days post-infection (dpi) and viral load (VL; area under the curve of log-PCR viremia from 0-21 dpi). Piglets were genotyped using the Illumina PorcineSNP60 Beadchip. Previous results using the NVSL trials of this data identified a large QTL on Sus scrofa chromosome (SSC) 4 that influenced both VL and WG. Genomic association analyses of VL and WG using all data and data from each PRRSV isolate separately were performed using Bayes-B. Genomic prediction accuracies were calculated as the correlation between genomic prediction and phenotype pre-adjusted for fixed effects, divided by square root of heritability. Marker-based heritabilities when analyzing all data were 0.43 for VL (0.47 for NVSL; 0.53 for KS06) and 0.35 for WG (0.36 for each isolate). The SSC4 QTL explained the largest amount of genetic variance for VL for both isolates (13.4% for NVSL; 7.5% for KS06) and also for WG in the NVSL trials, 10.8%. This QTL was not associated with WG in the KS06 trials (explained 0.11% of genetic variance), possibly because KS06 is less virulent, resulting in lower VL than the NVSL isolate. All other 1-Mb windows explained less than 2.2% (VL) and 3.6% (WG) of the genetic variance for both isolates and were not consistent between isolates. Accuracies of genomic prediction for KS06 data when training on NVSL were 0.32 (VL) and 0.40 (WG), respectively, while accuracies for NVSL data when training on KS06 were 0.42 (VL) and 0.17 (WG). When the SSC4 QTL genotype was included as a fixed-effect in the model and its correspondent 5-Mb region was removed from the genomic analyses, accuracies for VL based on the rest of the genome decreased to 0.14 for both training data analyses but remained unchanged for WG. These results show that there are genomic regions other than the SSC4 region that explain genetic variation in response to both PRRS isolates and that genomic prediction of host response to PRRSV infection for two isolates is possible. Support from PRRS-CAP, USDA-NIFA 2008-55620-19132, Genome Canada, and the breeding companies that provided pigs.

Keywords: PRRS, GWAS, prediction