I have been using both R and Mothur to do some of the statistic tests with my data. In R, with the function Adonis, there is not a way of doing pairwise comparisons and people suggest to just do independent adonis tests to pairs of your data.
In mothur, AMOVA (I understand equivalent of Adonis in R) gives you all pairwise comparisons. How are these calculated? Are they done through adonis too, but instead of having to modify your distance matrix independently (like in R), Mothur does it for you?
If so, I have read that when doing pairwise comparisons using ADONIS it is recommended to correct the p value using the Benjamini or Bonferroni corrections, among others. Does Mothur accounts for this. If it does not, do you have any recommendation of how to do it?
I tried p.adjust in R, but it does not make sense to me because if I am running adonis only between two groups (let’s say group A with 5 samples and group B with another 5 samples) I am only comparing group A and B. Therefore, in the p.adjust command in R, length n would be 1, and my p value would not change. I am probably doing something wrong.
In mothur, AMOVA (I understand equivalent of Adonis in R) gives you all pairwise comparisons. How are these calculated? Are they done through adonis too, but instead of having to modify your distance matrix independently (like in R), Mothur does it for you?
If the amova command gets a p-value below 0.05 it will automatically do the pairwise comparisons between the different groups by performing amova on the pairs. You don’t have to make the matrices, amova does this automatically. If your original p-value was greater than 0.05, then you will not get the pairwise comparisons.
If so, I have read that when doing pairwise comparisons using ADONIS it is recommended to correct the p value using the Benjamini or Bonferroni corrections, among others. Does Mothur accounts for this. If it does not, do you have any recommendation of how to do it?
I tried p.adjust in R, but it does not make sense to me because if I am running adonis only between two groups (let’s say group A with 5 samples and group B with another 5 samples) I am only comparing group A and B. Therefore, in the p.adjust command in R, length n would be 1, and my p value would not change. I am probably doing something wrong.
mothur doesn’t do any corrections, but will tell you what your p-values should be by Bonferonni. If you only have one comparison that you are interested in, then correcting for multiple comparisons doesn’t make sense.