Relative abundance calculation after tax_glom() gives non-representative relative abundance in figures about microshades

[lakarstens]

Thank you for pointing this out and the very detailed example! You are correct that the na's are dropped in the tax_glom function. We will work on making this more apparent to users and potentially disabling it.

In the meantime, the way around this issue is to perform all the functions that occur in prep_mdf independently rather than running the prep_mdf function (i.e. run tax_glom with NArm = FALSE, transform_sample_counts, and ps_melt).

Example

# perform custom mdf_prep
    mdf <- ps %>%

 # agglomerate at specified subgroup level (i.e. Genus)
    speedyseq::tax_glom(subgroup_level, NArm = FALSE) %>%

 # transform sample counts / normalize as desired
    phyloseq::transform_sample_counts(function(x) { x/sum(x) }) %>%

 # melt the phyloseq object
    speedyseq::psmelt()

 # Remove zeros
    mdf_prep <- mdf[mdf$Abundance > 0, ]

[sterrettJD]

Thanks for looking into the solution and sending the new prep_mdf code!

[KarstensLab]

Hi @sterrettJD,
The latest version of microshades now has the option to remove NAs and the default behavior is to NOT remove NAs. See this vignette for more details.

Thanks again for bringing this to our attention!
~Lisa