You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
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?
The text was updated successfully, but these errors were encountered:
I used run_chemrich_chemical_classes() on t-test results, inputting compound name, smiles, p-values, effect-size, and sub-pathway information as 'set'.
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.
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?
The text was updated successfully, but these errors were encountered: