Abstract

Improving fruit quality traits such as metabolic composition remains a challenge for tomato breeders. To better understand the genetic architecture of these traits and decipher the demographic history of the loci controlling tomato quality traits, we applied an innovative approach using multiple haplotype-based analyses, aiming to test the potentials of haplotype based study in association and genomic prediction studies. We performed and compared haplotype vs SNP-based associations (hapQTL) with multi-locus mixed model (MLMM), focusing on tomato fruit weight and metabolite contents (i.e. sugars, organic acids and amino acids). Using a panel of 163 tomato accessions genotyped with 5995 SNPs, we detected a total of 784 haplotype blocks, with an average size of haplotype blocks ~58 kb. A total of 108 significant associations for 26 traits were detected thanks to Haplotype/SNP-based Bayes models. Haplotype-based Bayes model (97 associations) outperformed SNP-based Bayes model (50 associations) and MLMM (53 associations) in identifying marker-trait associations as well as in genomic prediction (especially for those traits with moderate to low heritability). To decipher the demographic history, we identified 24 positive selective sweeps using the integrated haplotype score (iHS). Most of the significant associations for tomato quality traits were located within selective sweeps (54.63% and 71.7% in hapQTL and MLMM models, respectively). Promising candidate genes were identified controlling tomato fruit weight and metabolite contents. We thus demonstrated the benefits of using haplotypes for evolutionary and genetic studies, providing novel insights into tomato quality improvement and breeding history.

中文翻译：

---

基于多个单倍型的分析提供了对番茄果实重量和组成的遗传和进化见解

 抽象的

改善代谢组成等果实品质性状仍然是番茄育种者面临的挑战。为了更好地了解这些性状的遗传结构并破译控制番茄品质性状的基因座的人口统计历史，我们应用了一种使用基于多个单倍型分析的创新方法，旨在测试基于单倍型的研究在关联和基因组预测研究中的潜力。我们使用多位点混合模型 (MLMM) 进行并比较了单倍型与基于 SNP 的关联 (hapQTL)，重点关注番茄果实重量和代谢物含量（即糖、有机酸和氨基酸）。使用具有 5995 个 SNP 基因分型的 163 个番茄种质，我们总共检测到 784 个单倍型块，单倍型块的平均大小约为 58 kb。通过基于单倍型/SNP 的贝叶斯模型，总共检测到 26 个性状的 108 个显着关联。在识别标记-性状关联以及基因组预测（特别是对于那些具有中度至低遗传力的性状）方面，基于单倍型的贝叶斯模型（97 个关联）优于基于 SNP 的贝叶斯模型（50 个关联）和 MLMM（53 个关联）。为了破译人口统计历史，我们使用综合单倍型评分 (iHS) 确定了 24 个阳性选择性扫描。大多数番茄品质性状的显着关联都位于选择性扫描中（在 hapQTL 和 MLMM 模型中分别为 54.63% 和 71.7%）。确定了控制番茄果实重量和代谢物含量的有希望的候选基因。因此，我们证明了使用单倍型进行进化和遗传研究的好处，为番茄品质改良和育种历史提供了新的见解。