The summary plot doesn’t generate any significant values (asterisks) in the plot, but the significant_results.tsv table shows the significance.
Does this mean the model doesn’t fit my data? How is pFDR calculated?
Would you mind providing a chunk of the significance table and plot so I can see what you mean? (You can obscure species and metadata names if necessary.) Nothing seems wrong with the command itself at first glance.
Unless the model errors out, it has fit the data and should show up in both the plot and results file if significant. pFDR for the plot is from the qval_individual column of the significant results. Is that the column you’re looking at when you say the significant_results.tsv table shows significance?
Is it necessary to send the association plot as well?
Yes, I looked at the qval_individual column in the significant_results.tsv table, which shows the significance. However, I don’t see the same representation in the plot.
My input abundance table was extract from feature table of phyloseq object. Is it correct? Is some normalization necessary before entering into Maaslin3?
The two covariates that have significant associations in the table (sample_size and Outcome) are both shown on the coefficient plot side with coloring by significance. The legend shows that the shading goes to 1e-4, which seems to match your results table. The bottom of the plot is cut off, but presumably the heatmap variable is Collection_time. I don’t see any significant associations with Collection_time in the table, so it makes sense that there are no stars on the heatmap size of the plot (stars are only for the heatmap side). If you want to force sample_size and Outcome no_sepsis to show up in the heatmap side as well (where they’ll have significance stars), you can set heatmap_vars = c("sample_size", "Outcome no_sepsis").
