简单回归尺度转换实现半滑舌鳎性逆转基因的高效定位

在间断性状全基因组关联分析中,当基因组数据存在复杂群体分层时,广义线性模型需要同时考虑上百个协变量,其求解速度会大大下降而且还会产生异常解.本研究目的是把简单回归结果中显著位点的效应值和遗传力的尺度转化为可解释的广义线性回归结果.首先对亲缘关系矩阵进行谱分解,特征向量作为主成分(PC),矫正间断性状中的群体分层;再求解...

Efficiently mapping the sex reversal genes of half-smooth tongue sole, Cynoglossus semilaevis using simple regression scale transformation

In genome-wide association analysis of discontinuous traits, when complex population stratification exists in genomic data, the generalized linear model needs to consider hundreds of covariables at the same time, which slows the calculation speed and presents abnormal solutions. This study aimed to transform the effect value and heritability scale of significant loci in simple linear regression results into interpretable generalized linear regression results. First, the eigenvectors solved by spectral decomposition of the kinship matrix were considered as the principal components (PCs) to correct the population stratification in the discontinuous traits dataset. Then, a new covariate was formed through the sum of the multiplications of each covariate, and its regression coefficient of the principal component was computed using a linear regression model. The new covariate was used as the covariable of simple regression to carry out correlation tests for markers one by one. Finally, the generalized linear model was used for regression analysis of candidate quantitative trait nucleotides (QTNs), and the effects and variance were transformed into the generalized linear regression model scale. The genome-wide association analysis of sex reversal traits in half-smooth tongue sole (Cynoglossus semilaevis) was conducted using the new method and the generalized linear regression model with direct consideration of principal components: The results show that the QTN detection efficiency of this method is higher, a total of 6 QTNs were detected, including 5 QTNs on Z chromosome and 1 QTN on W chromosome. In addition, in terms of genome control, the genome control value of the method in this study is the same as that of the generalized linear regression model which directly considers PC, which is at an optimal level of 1.01. Therefore, the simple regression scaling transformation method based on principal component analysis improved the detection power for QTN detection, while retaining the accuracy of results, with fast and robust genome-wide association analysis of discontinuous traits. In addition, the QTNs detected by the new method proposed in this study can provide theoretical guidance for the study of sex reversal traits in half-smooth tongue soles.