comparing different variance between samples (amplicon data)

Question

[atenasadat.sombolestani]

Hi everyone,

I'm analyzing my amplicon data, 16S and ITS, and faced a problem for analyzing the possible correlation between my variables. I have six variables (in my metadata table) and want to know if the variance of my ASVs between two groups are meaningful or not. For example I want to know if the variance between variable 1 and variable 2 are statistically meaningful or not (p value< 0.05). I first did adonis(distance_matrix~ variable 1 *variable 2) and got p value and R square values for my two variables separately and together, but I realized that the order of variables matter on one hand and on the hand I don't get the difference between this piece of code and this one adonis(distance_matrix~ variable 1 + variable 2)! My data are not normally distributed by the way. Does anyone know any solution? Thanks in advance

Answer 1

[Asaf]

The difference between ~ Var1 * Var2 and ~ Var1 + Var2 is simply the interaction between Var1 and Var2 being a part of the design equation in the first case and omitted in the second case. Var1 * Var2 is short for Var1 + Var2 + Var1:Var2. The interaction term tests the hypothesis that the effect of Var2 depends on the value of Var1.