I have encountered an inconsistency in the LEfSe module outputs in MicrobiomeAnalyst.
The graphical summary does not appear to correspond to the taxa listed in the results table. In addition, the results table includes taxa marked as significant even though their LDA scores fall below the threshold I specified (log LDA ≥ 3.0).
This raises two questions:
Is this a known issue related to the display or filtering logic in the LEfSe module?
Should the results table be interpreted independently of the LDA threshold applied in the graphical outputs?
At present, it seems necessary to export the table and manually filter taxa based on both FDR and LDA criteria before generating plots. I would appreciate confirmation on whether this is the recommended approach.
So, I applied filtering criteria → on the first sight graphical summary looks fine (as LDA really is above my threshold). Then I switch to results table – it turns out that previously listed features not necesserily have FDR adjusted p <0.05, while highlighted features listed in the table do not have LDA > |2.5| what I have indicated.
I see the issue. MicrobiomeAnalyst highlights the top sig. rows by p-values, not necessarily according to LDA when the table is the same. You can either
Wait for the next server update (~3 days)
Force the table to be regenerated (i.e. switching to a different tax. level, then change it back)
