Different pathway analysis results between MetaboAnalystR and Web platform

MetaboAnalyst
, ,
1

Hello!
When using the MetaboAnalyst 6.0 Pathway Analysis module, using a list of compound names and a self-specified reference metabolome, I get the following results:

pathway_impact_reference_metabolome
pathway_impact_reference_metabolome504×504 17.9 KB

and the list of impacted pathways is led by Glycolysis / Gluconeogenesis (hsa00010), 13, 4.67, 9, 1.06E-02, 1.98E+00, 5.92E-01, 3.01E-01, 0.47.

However, using MetaboAnalystR, using the same settings (relative betweenness centrality, hypergeometric test) and setting the metabolome filter using the same list of KEGG compounds as previously (using `mSet<-SetMetabolomeFilter(mSet, T)

mSet<-Setup.KEGGReferenceMetabolome(mSet, “data//KEGG_list_PCDL.txt”)`)
I get different results.
The pathway impact graph looks like this:

path_view_1_dpi72
path_view_1_dpi72504×504 17.2 KB

and whilst the top result for impacted pathways is the same, the rest are different, and the results for Total/Expected/Hits/Raw p value/-log10p/Holm adjust/FDR/Impact are all different.

What could the reasons for this be, and which set of results should I trust?

Thank you!

2

Hi keiko,

I recently encountered a similar issue.

Have you compared the MetaboR commands given on the website and/or on the download report with your locally run version of the “same” R script?

In my case some choices I had made on the website did not show up in the R script in the report or in the R commands. After I adjusted for that, the results were similiar again.

As to why this happened/happens I can’t tell you.

Cheers,
MSC