Hello!
Thanks so much for making Maaslin3! I am really liking it so far, but had a question about a discrepancy i found within some of my outputs.
Not all of the line graphs that are produced and labeled with a significant q-value are within the significant_results.tsv. For example, see the Maaslin3 produced graph i’ve linked below (it’s a weird looking one but also the reason i used as an example). This particular OTU is shown to have an FDR qval of 3.1e-02. However, when i find it in the ‘all_results.tsv’ this is the info given for it, not exactly lining up. It is not for all of them, as some of them do indeed match up between graphs and text files in terms of qvalues – so is it because of some correction, etc that the qvalues are changing between the produced graphs and the text files?
Sorry, i hope this makes sense! Thanks for your help in advance!
| feature | metadata | value | name | coef | null_hypothesis | stderr | pval_individual | qval_individual | pval_joint | qval_joint | error | model | N | N_not_zero |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OTU025 | AST | AST | AST | -1.3981705 | 0.05880495 | 0.40257938 | 0.00043592 | 0.08785455 | 0.00043592 | 0.06795249 | NA | abundance | 247 | 205 |
My maaslin command:
fit_out ← maaslin3(
input_data = bacteria,
input_metadata = metadata,
output = output_dir,
normalization = “NONE”, ##Samples are already TSS
transform = “LOG”,
formula = "~ Treatment + AST + AST:Treatment + (1|Batch),
augment = TRUE,
standardize = TRUE,
max_significance = 0.05,
median_comparison_abundance = TRUE,
min_abundance = 0.0001,
min_prevalence = 0.1,
max_pngs = 100,
save_models = TRUE
)
. This was one of those graphs. If i look at just the ones that were made with this most recent run, they are correct.