Effects of marker density and population structure on the genomic prediction accuracy for growth trait in Pacific white shrimp Litopenaeus vannamei

• Main Authors: Quanchao Wang, Yang Yu, Jianbo Yuan, Xiaojun Zhang, Hao Huang, Fuhua Li, Jianhai Xiang
• Format: Article
• Language: English
• Published: BMC 2017-05-01
• Series: BMC Genetics
• Subjects: Heritability; Genomic selection; Growth traits; Penaeid shrimp
• Online Access: http://link.springer.com/article/10.1186/s12863-017-0507-5

Abstract Background Due to the great advantages in selection accuracy and efficiency, genomic selection (GS) has been widely studied in livestock, crop and aquatic animals. Our previous study based on one full-sib family of Litopenaeus vannamei (L. vannamei) showed that GS was feasible in penaeid shrimp. However, the applicability of GS might be influenced by many factors including heritability, marker density and population structure etc. Therefore it is necessary to evaluate the major factors affecting the prediction ability of GS in shrimp. The aim of this study was to evaluate the factors influencing the GS accuracy for growth traits in L. vannamei. Genotype and phenotype data of 200 individuals from 13 full-sib families were used for this analysis. Results In the present study, the heritability of growth traits in L. vannamei was estimated firstly based on the full set of markers (23 K). It was 0.321 for body weight and 0.452 for body length. The estimated heritability increased rapidly with the increase of the marker density from 0.05 K to 3.2 K, and then it tended to be stable for both traits. For genomic prediction on the growth traits in L. vannamei, three statistic models (RR-BLUP, BayesA and Bayesian LASSO) showed similar performance for the prediction accuracy of genomic estimated breeding value (GEBV). The prediction accuracy was improved with the increasing of marker density. However, the marker density would bring a weak effect on the prediction accuracy after the marker number reached 3.2 K. In addition, the genetic relationship between reference and validation population could influence the GS accuracy significantly. A distant genetic relationship between reference and validation population resulted in a poor performance of genomic prediction for growth traits in L. vannamei. Conclusions For the growth traits with moderate or high heritability, such as body weight and body length, the number of about 3.2 K SNPs distributed evenly along the genome was able to satisfy the need for accurate GS prediction in the investigated L.vannamei population. The genetic relationship between the reference population and the validation population showed significant effects on the accuracy for genomic prediction. Therefore it is very important to optimize the design of the reference population when applying GS to shrimp breeding.