Abstract Streptococcosis is a major disease that causes severe mortality in tilapia aquaculture worldwide. Although the conventional BLUP family selection to enhance disease resistance in a commercial red tilapia stock was successful, the response was low due to the low heritability of the traits. An alternative strategy is the utilization of genomic information to identify the best performing candidates within families. In this study, we performed genome-wide association studies for red tilapia resistance to Streptococcus agalactiae using 11,480 SNPs within 110 families represented by 1020 fish. Nineteen SNP markers were found to explain ~10% of the genetic variation. We compared the accuracies of genomic prediction using the pedigree-based (PBLUP), marker-based (GBLUP), and Bayesian models. The prediction accuracy was assessed by performing ten replicates of five-fold cross-validation. In each replicate, approximately 80% of the data (n~816) were sampled for the training set and the remaining data (n~204) were used for the validation. The BayesB model yielded the highest accuracies (0.31 and 0.20) followed by GBLUP (0.25 and 0.15) and PBLUP (0.15 and 0.06) for days to death and binary trait.

中文翻译：

---

杂交红罗非鱼（Oreochromis spp.）对无乳链球菌抗性的全基因组关联研究和基因组预测

摘要 链球菌病是导致全球罗非鱼养殖严重死亡的主要疾病。尽管用于增强商业红罗非鱼种群抗病性的常规 BLUP 家族选择是成功的，但由于性状的低遗传性，反应较低。另一种策略是利用基因组信息来确定家族中表现最佳的候选者。在这项研究中，我们使用 1020 条鱼代表的 110 个家族中的 11,480 个 SNP，对红罗非鱼对无乳链球菌的抗性进行了全基因组关联研究。发现 19 个 SNP 标记可以解释约 10% 的遗传变异。我们使用基于谱系 (PBLUP)、基于标记 (GBLUP) 和贝叶斯模型比较了基因组预测的准确性。通过执行十次重复的五倍交叉验证来评估预测准确性。在每个重复中，大约 80% 的数据 (n~816) 被采样用于训练集，其余数据 (n~204) 用于验证。BayesB 模型的准确度最高（0.31 和 0.20），其次是 GBLUP（0.25 和 0.15）和 PBLUP（0.15 和 0.06）对于死亡天数和二元性状。