Russel88
commented
3 years ago Hi Mia,
The warning does not tell you whether or not there are any differences between the phyla. The problem is that testDA needs enough features (here phyla) to actually compare the different DA methods. With very few features the DA methods cannot be reliably compared.
You could run allDA and compare the results of the different methods:
test_all <- allDA(p1, predictor = "Species")
test_all$adj
test_all$est
or simply choose a DA method that usually works well and go with that, such as DA.wil, DA.ds2, or DA.msf
Regards, Jakob
Hello,
I want to test the difference between two species at the Phylum level. Based on their relative abundance plots, I do not expect them to be different. When I try to run DA test at the phylum level I get the following error:
Set to spike all features. Can't calculate FPR or AUC. Change k argument In addition: Warning message: In testDA(p1, predictor = "Species") : Dataset contains very few features
If I try this at Class or a different taxonomic level, I don't get the error. Should I assume there aren't differences at the Phylum level?
Thank you, Mia