Génétique Animale et Biologie Intégrative, INRA, AgroParisTech, Université Paris-Saclay, Jouy-en-Josas, France.
Center for Quantitative Genetics and Genomics, Aarhus University, Tjele, Denmark.
PLoS One. 2018 Jul 2;13(7):e0199931. doi: 10.1371/journal.pone.0199931. eCollection 2018.
Using GWAS to identify candidate genes associated with cattle morphology traits at a functional level is challenging. The main difficulty of identifying candidate genes and gene interactions associated with such complex traits is the long-range linkage disequilibrium (LD) phenomenon reported widely in dairy cattle. Systems biology approaches, such as combining the Association Weight Matrix (AWM) with a Partial Correlation in an Information Theory (PCIT) algorithm, can assist in overcoming this LD. Used in a multi-breed and multi-phenotype context, the AWM-PCIT could aid in identifying udder traits candidate genes and gene networks with regulatory and functional significance. This study aims to use the AWM-PCIT algorithm as a post-GWAS analysis tool with the goal of identifying candidate genes underlying udder morphology. We used data from 78,440 dairy cows from three breeds and with own phenotypes for five udder morphology traits, five production traits, somatic cell score and clinical mastitis. Cows were genotyped with medium (50k) or low-density (7 to 10k) chips and imputed to 50k. We performed a within breed and trait GWAS. The GWAS showed 9,830 significant SNP across the genome (p < 0.05). Five thousand and ten SNP did not map a gene, and 4,820 SNP were within 10-kb of a gene. After accounting for 1SNP:1gene, 3,651 SNP were within 10-kb of a gene (set1), and 2,673 significant SNP were further than 10-kb of a gene (set2). The two SNP sets formed 6,324 SNP matrix, which was fitted in an AWM-PCIT considering udder depth/ development as the key trait resulting in 1,013 genes associated with udder morphology, mastitis and production phenotypes. The AWM-PCIT detected ten potential candidate genes for udder related traits: ESR1, FGF2, FGFR2, GLI2, IQGAP3, PGR, PRLR, RREB1, BTRC, and TGFBR2.
使用 GWAS 从功能层面鉴定与牛体貌特征相关的候选基因具有挑战性。在奶牛中广泛报道的长程连锁不平衡 (LD) 现象是鉴定与这种复杂特征相关的候选基因和基因相互作用的主要难点。系统生物学方法，如将关联权重矩阵 (AWM) 与信息理论中的偏相关 (PCIT) 算法相结合，可以帮助克服这种 LD。在多品种和多表型背景下，AWM-PCIT 可以帮助鉴定具有调节和功能意义的乳房特征候选基因和基因网络。本研究旨在使用 AWM-PCIT 算法作为 GWAS 后的分析工具，以鉴定乳房形态的候选基因。我们使用了来自三个品种的 78440 头奶牛的数据，这些奶牛具有五个乳房形态特征、五个生产特征、体细胞评分和临床乳腺炎的自身表型。奶牛使用中密度（50k）或低密度（7 到 10k）芯片进行基因分型，并被 imputed 到 50k。我们在品种内和特征内进行了 GWAS。GWAS 在整个基因组中显示了 9830 个显著 SNP（p<0.05）。5010 个 SNP 未映射到基因，4820 个 SNP 在基因的 10-kb 内。在考虑到 1SNP:1gene 后，3651 个 SNP 在基因的 10-kb 内（set1），并且 2673 个显著 SNP 距离基因 10-kb 以上（set2）。这两个 SNP 集形成了 6324 个 SNP 矩阵，该矩阵在 AWM-PCIT 中进行拟合，考虑到乳房深度/发育作为关键特征，导致 1013 个与乳房形态、乳腺炎和生产表型相关的基因。AWM-PCIT 检测到 10 个与乳房相关特征相关的潜在候选基因：ESR1、FGF2、FGFR2、GLI2、IQGAP3、PGR、PRLR、RREB1、BTRC 和 TGFBR2。
