I have seen a similar question asked before, but I am not entirely sure about the answer. I want to perform a LEfSe analysis, and the data I have does not contain a subclass. I am noticing that upon running LEfSe, I see the message “Number of significantly discriminative features: 0 ( 17 ) before internal wilcoxon.” However, this changes when I change the Alpha value for the wilcoxon test between subclasses from 0 to 1, and then I see significant features. I thought that the program would not perform a subclass wilcoxon test if there were no subclasses, and so that alpha value would not matter. My question is that if I do not want a subclass wilcoxon test to be performed, should I change the Alpha value to 1, so that there is no filtering based off subclass? Or does the program skip the wilcoxon test by default if I specified there were no subclasses in the previous step?
Thanks in advance!