metastats "nan" errors

Is metastets capable of comparing phylotypes across sets that have only a single group in them?

Here is my design file with 4 groups divided into 2 sets:
W1 W
W2 W
W3 W
Sed S

mothur > metastats(,, sets=W-S, groups=Sed-W1-W2-W3)

Metastats runs fine but the output file ( looks like: OTU mean(group1) variance(group1) stderr(group1) mean_of_counts(group1) mean(group2) variance(group2) stderr(group2) mean_of_counts(group1) p-value 1 nan -0.000000000000 nan nan 2057.692965960159 2722742.770169086289 825.036782538980 2874.750000000000 1.000000000000 2 nan -0.000000000000 nan nan 243.359057971014 236894.524386158359 243.359057971014 250.250000000000 1.000000000000 3 nan -0.000000000000 nan nan 174.079493138192 101769.257845592918 159.506471534538 189.500000000000 1.000000000000 4 nan -0.000000000000 nan nan 152.579699706629 50466.524147885713 112.323777700767 163.500000000000 1.000000000000

It looks like the mean of group 1 (S) cannot be calculated. However, that doesn’t make much sense. If there is only a single incidence of a phylotype in the group, then the mean should just end up being that value. I don’t understand how “nan” could be appearing.

For example, if Phylotype 1 has an incidence of 200 in Sed then the mean of its incidence across group S should be 200. Am I wrong here? Why is “nan” appearing?

Does this mean that it is not possible to use metastats when you’d like to compare phylotypes across sets that have a single group? I found that if I change the sets so that there are 2 groups in each set then metastats works just fine.


No, sorry that won’t work. metastats is basically a t-test. This means that you have to have replicates within a treatment to measure the variation within a treatment.


Good to know. I’ll keep that in mind.