基于偏最小二乘路径模型的整体基因与多数量性状交互作用的统计推断

目的 用偏最小二乘路径模型(PLSPM)推断整体基因对多数量性状的交互作用。方法 通过数据模拟的方法评价基于PLSPM统计量的稳定性与检验效能,并对实例数据中的FTO和NEGR1两个基因进行分析,研究它们对相关多数量性状的交互作用。结果 统计模拟结果显示,基于PLSPM提出的检验统计量一类错误稳定,检验效能随着样本量的增加以及效应值的增大而升高。实例数据分析表明,基因区域FTO (Chr16:52390317.. 52391106)和NEGR1(Chr1:71707640..71709688)对相关多数量性状的交互作用有统计学意义(P=0039)。结论 对于两个整体基因对多数量性状的交互作用研究,提出的基于PLSPM统计量是一种有效的分析方法。

Statistical inference for detecting gene-gene interaction on multiple quantitative traits based on partial least squares path model

Objective To infer gene-gene interaction on multiple quantitative traits using partial least squares path model (PLSPM). Methods Simulation was used to evaluate the stability and power of the PLSPM-based statistic, which was further applied to identify the interaction between FTO gene and NEGR1 gene on related multiple quantitative traits in one real dataset. Results The proposed PLSPM-based statistic was stable and its power increased with the increasing of sample size and interaction effect. The interaction of gene regions FTO (Chr16:52390317.. 52391106) and NEGR1 (Chr1:71707640.. 71709688) for correlated multiple quantitative traits showed statistical significance (P=0039). Conclusion The proposed PLSPM-based statistic is powerful to detect gene-gene interaction for multiple quantitative traits.