冠心病循环miR-92a数据的正态变换方法比较

目的分析冠心病循环miR-92a表达量数据分布特点,探讨其正态变换方法。方法比较对数、Box-Cox及Johnson三种正态变换方法对冠心病循环miR-92a表达量数据变换效果。结果冠心病循环miR-92a表达量数据变异性强烈,呈正偏态分布(偏度=10.90)。对数变换无法使冠心病循环miR-92a表达量数据服从正态分布(KS值=0.071,P=0.002)。Box-Cox变换虽使数据的正偏度明显减少,但仍未能通过Kol-mogorov-Smirnov正态检验(KS值=0.060,P=0.021)。Johnson变换结果最优,其通过了K-S检验(KS值=0.049,P=0.200)。结论 Johnson变换具有更强的正态变换能力,是变异性强的冠心病循环miR-92a表达量数据正态变换的理想工具。

Comparison of different normal transformation methods in data analysis of circulating mir-92a in patients with coronary artery disease

【Objectives】 To analyze data distribution of microRNA-92a in patients with coronary artery disease and to investigate normalization method of the data.【Methods】 To compare the effect of logarithm transformation,BOX-COX transformation and Johnson transformation on circulating miR-92a in patients.【Results】 Circulating mir-92a levels are not normal distribution.It cannot be normalized by logarithm transformation(KS value=0.071,P =0.002).The level of normality of data of miR-92a can be highly more promoted through Johnson transformation(KS value=0.049,P =0.200) than through Box-Cox transformation(KS value=0.060,P =0.021).【Conclusion】 Johnson transformation is the best way for circulating miR-92a levels.