基于组LASSO的重组近亲杂交合作小鼠基因位点定位研究

重组近亲杂交合作小鼠品系(CC-RIX)具有很多优点,特别是在复杂疾病的数量性状位点定位方面,具有较高功效.本文在现有研究基础上,考虑了仅含主基因效应的混合线性定位模型,并通过组LASSO惩罚函数法对问题进行转换,然后采用迭代加权最小二乘法对转换后的问题进行求解.从而克服了设计矩阵容易奇异,计算速度慢等计算上的难题.模拟计算表明,本文所提模型和方法在CC-RIX品系的数量性状位点定位中能快速、准确地识别出数量性状位点,并具有较高的真阳性率,以及较低的假阳性率.

Applying Group LASSO to Map the Quantitative Trait Loci for Recombinant Inbred Intercrosses Lines Derived From Collaborative Cross Mice

The recombinant inbred intercrosses derived from the collaborative cross(CC-RIX)have many merits,especially the higher mapping power in quantitative trait loci(QTL)for complex disease.Based on the existing studies,the paper establishes a linear mixed model that only includes the main founder effects to map the QTLs.The group LASSO penalized method has been applied to solve the model,and the iteratively re-weighted least-squares algorithm was used to find the optimal estimation.By doing that,some problems,such as the singularity of the design matrix,the computational cost,and so on,have been overcome.The simulation results show that our proposed method can identify the QTLs fastly and correctly for CC-RIX lines with a high true positive rate(TPR)and low false-positive rate(FPR).