Analysis of Variance Heterogeneity in Metabolic Traits Among Full CSS Mice
Mao Ding
2025-01-28
☆ Objective
This study aims to assess and compare the variance and mean effects of metabolic traits between every CSS strain and B6 mice. The metabolic traits include body length, body weight gain, BMI, glucose, insulin, HOMA(Homeostatic model assessment, a measurement used to quantify insulin resistance and beta-cell function), Cholesterol, TG(triglycerides, either in plasma or in liver), liver weight, liver TG, and total liver TG. The null hypothesis is none of the CSS strains mice will differ significantly from the B6 reference strain for any trait, whether considering central tendency or dispersion.
☆ Density Plot for BMI
☆ Skewness & Kurtosis
☆ Distribution Tests
Three different methods were employed to assess the distribution of the data. The tests include:
Jarque-Bera Normality Test: This test evaluates whether a sample’s skewness and kurtosis match those of a normal distribution, helping to determine if the data follows a normal distribution.
Kolmogorov-Smirnov (KS) Test: This test checks for normal and weibull distributions. It calculates the maximum distance between the empirical cumulative distribution function (ECDF) of the sample data and the cumulative distribution function (CDF) of the normal or weibull distribution. This test is sensitive to sample size, with larger samples, even minor deviations from normality can yield significant results. It may be less powerful than other normality tests for certain types of non-normal distributions.
Anderson-Darling (AD) Test: similar to the KS test, the AD test evaluates normal and weibull distributions. It calculates a test statistic (A²) based on the difference between the ECDF of the sample and the CDF of the normal or weibull distribution. The AD test is generally regarded as more powerful than the KS test, particularly for detecting differences in the tails of the distributions.
Null Hypothesis (H0): The data follows a normal or weibull distribution.
Alternative Hypothesis (H1): The data does not follow a normal or weibull distribution.
Cells highlighted in pink indicate that the p-value for the corresponding distribution test is greater than 0.05, suggesting that the data follows a normal or weibull distribution.