I am trying to generate PERMANOVAS using the adonis function from the vegan package. I am testing many different factors to see which ones are significant. For certain factors in my metadata some samples are NA. Obviously when I try to generate a permanova for these factors, I get an error. I don't want to remove rows that have NA values from the dataset because they are only in certain columns, and I would like to include these samples for analysis of other factors. I also do not want to fill the NA columns with another value (ie 0) because this may effect the results. I am looking for a way to exclude the values that are NA when generating each permanova.
I tried both
propData<-as.data.frame(t(otu_table(otu)))
propData_bray<-vegdist(propData, "bray")
adonis(formula = propData_bray ~ pH, data=new_meta, na.action=na.exclude)
and
adonis(formula = propData_bray ~ pH, data=new_meta, na.action=na.omit)
The otu table has no NA values. Here is an example of the metadata:
| sample | factorA | factorB | factorC | factorD |
|---|---|---|---|---|
| 1 | yes | 3 | NA | 0.5 |
| 2 | yes | 4 | no | 1 |
| 3 | no | NA | yes | 0.05 |
| 4 | yes | 3 | yes | 0.7 |
| 5 | NA | 0 | NA | 1.5 |
for factorA I would include samples 1-4, factorB 1-2 & 4-5 , etc.