[目的]运用多因子降维法(MDR)分析中国荷斯坦牛CXCR1和IL-8两个基因多态性与乳房炎易感性的交互作用,并评价MDR模型的具体应用.[方法]以上海光明乳业集团金山种奶牛场634头(其中有228头患乳房炎)中国荷斯坦牛为试验材料,采用PCR-SSCP和直接测序法分析CXCRl基因-5'端和编码区及IL-8基因编码区共7个SNP位点的遗传多态性,并用MDR法和Logistic回归分析CXCR1和IL-8基因各位点突变及其基因型组合对奶牛乳房炎易感性的影响.[结果]CXCR1基因5'端-1 830、-1 768、-344和编码区783共4处发生突变,IL-8基因的2 789(A/G)位点与2 862(T/C)位点发生连锁突变,并导致两位点分别发生Ser→Gly和Cys→Arg的突变.单因素Logistic回归分析结果表明,CXCR1基因和IL-8基因5个SNP位点变异对该牛场奶牛乳房炎易感性均没有达到显著水平,但MDR分析发现CXCR1-(-1 768)-IL-8两位点、CXCR1-(-1 768)-(783)-IL-8三位点及CXCR1-(-1 768)-(-344)-(783)-IL-8四位点交互作用均达到显著水平,且交叉验证一致性均为最大(10/10).CXCR1-(-1 768)-(78 3)-IL-8三位点互作模型和CXCR1-(-1 768)-IL-8两位点互作模型检验样本准确度较高(分别为0.5457和0.5485),但4个位点联合效应分级图表明CXCR1-(-1 768)-IL-8两易感基因型为强正交互作用,同时将CXCR1-(-1 768)-(78 3)-IL-8三位点互作效应和CXCR1-(-1 768)-IL-8两位点互作效应引入Logistic回归分析表明,CXCR1-(-1 768)-IL-8两位点互作效应对该牛场奶牛乳房炎易感性达到显著水平.[结论]CXCR1-(-1 768)-IL-8两位点交互模型作为奶牛乳房炎易感性相关的基因联合选择模型最佳,MDR和Logistic回归均可用于奶牛乳房炎易感性多基因分析模型,MDR发现交互作用的能力比Logistic回归强,但两者有着各自的适用条件和优点.遇到有交互作用分析要求时,二者的联合应用效果更佳.%[Objective]Multifactor Dimensionality Reduction (MDR) model was used to identify the interaction of CXCR1 and IL-8 and their impacts on the risk of mastitis susceptibility of Chinese Holstein, and to test it's feasibility in analyzing genes-genes interactions.[Method]The polymorphisms of 7 SNP sites including the 5' flanking region and coding region of CXCR1 and exon 4 and exon 5 of IL-8 were detected by PCR-SSCP and direct sequencing for 634 Chinese Holstein, including 228 individual mastitis cow, from Jinshan Breeding Dairy Farm of Bright Dairy Group of Shanghai, and to identify the interaction of CXCR1 and IL-8 and their impacts on the risk ofmastitis susceptibility using MDR and Logistic regression.[Result]Four SNPs were found for CXCR1 including three SNPs (-1 830(A/G), -1 768(T/A), -344(T/C)) in 5' flanking region and 783(C/A) in coding region.A linked mutation in 2 789(A/G) and 2 862(T/C) sites was found for IL-8 gene, and this mutation results in the transaction of AA from Ser to Gly, and Cys to Arg.The main effects of polymorphism of 5 SNPs of CXCR1 and IL-8 gene showed no significant impacts on the mastitis susceptibility of cow based on the single factor Logistic regression.The interactions of CXCR1-(-1 768)-IL-8,CXCR1-(-1 768)-(783)-IL-8, and CXCR1-(-1 768)-(-344)-(783)-IL-8 showed significant impacts on the mastitis susceptibility of cow using MDR model, and the cross-validation consistency was maximum (10/10).The interactions of CXCR1-(-1 768)-IL-8 and CXCR1-(-1 768)-(783)-IL-8 showed high testing balance accuracy (0.5457 and 0.5485, respectively).The interaction of CXCR1(-1 768)-IL-8 was high degree synergy for mastitis susceptibility from the interaction dendrogram of multi-genotypes of CXCR1 and IL-8.In the meanwhile, the interactions of CXCR1-(-1 768)-IL-8 showed significant impacts on the mastitis susceptibility of cow using Logistic regression model including the main effect of 5 SNPs and the interaction effect of CXCR1-(-1 768)-IL-8 and CXCR1-(- 1 768)-(783)-IL-8 (P＜0.05).[Conclusion]The interactions of CXCR1-(- 1 768)-IL-8 was the best multi-genotype model to estimate the risk of mastitis susceptibility of Chinese Holstein.MDR and Logistic regression can also be used to estimate the risk of mastitis susceptibility of Chinese Holstein.MDR appeares better ability than Logistic regression for detection of high-order interactions, and MDR and Logistic regression have advantages and disadvantages when compared to each other.Combined use of MDR and Logistic regression may be preferable when dealing with interaction analysis.
展开▼
