Some significant chemical clusters omitted in impact plots

[JG1ANDONLY]

I noticed there were fewer significant (p-values smaller than 0.05) clusters displayed in the impact plot than those in the excel file. For example, in my analysis, there are 25 significant clusters in the generated excel file, but there are only 21 clusters in the impact plot displayed. Is there a specific standard for dropping certain clusters besides looking at p-values only?

[barupal]

Which chemrich function are you using ?

[JG1ANDONLY]

Which chemrich function are you using ?

I used run_chemrich_chemical_classes() on t-test results, inputting compound name, smiles, p-values, effect-size, and sub-pathway information as 'set'.

[barupal]

This line has the filtering criteria :
clustdf.alt.impact <- clustdf[which(clustdf$pvalues<0.05 & clustdf$csize>1 & clustdf$alteredMetabolites>1) ,]

cluster p-value should be < 0.05, cluster size should be at least 2 and significant metabolites within a cluster should be at least 2. This might explain the difference.

[JG1ANDONLY]

I see. But I think for some reason clusters whose p-values = 0.05 were also included in my result. Thank you!