Michael McLaren

If there is 1 too many samples to display, then "variable" is used instead of "sample"; and it also seems that the claimed number of additional samples can be off....

Give messages when samples or taxa are dropped

1

Examples of cases where it seems worth it to `message()` the user when samples or taxa are dropped include * Overwriting sample data with `sample_data(ps)

Apply consistent rules about allowed sample names?

2

There seems to be inconsistent handling of sample names by various phyloseq methods. For example, ```r x s1 a 7 #> 2 b 8 #> 3 c 9 sample_data(y) %>%...

Currently, no warning is given for typos such as `subset_taxa(ps, phylum = "Bacteroidetes")`. Instead, this command will simply run and not filter out anything. Contrast with `dplyr::filter()`, which gives a...

Suppose one wants to create bootstrap replicates of the phyloseq dataset. Doing ```r smpls

Currently if one performs a `subset_samples` operation that removes all samples, they get an error message ``` Error in validObject(.Object) : invalid class “sample_data” object: Sample Data must have non-zero...

To serve same purpose as in `tax_glom()`, to allow user to specify other values that should be treated like `NA`. Only needed for case of `tax_adjust = 2`.

E.g. the user could call ``` r psmelt(ps, c("taxon", "sample", "abundance")) ``` to name the first three columns ("taxon", "sample", "abundance") instead of the default ("OTU", "Sample", "Abundance").

(I think) `subset_samples()` and `subset_taxa()` will drop samples and taxa where the subsetting condition evaluates to NA. This is consistent with base R's `subset()` and dplyr's `filter()`, but is can...