Reproducibility Issue with Metfrag Results

[zmahnoor14]

Hello,

I have been working on the workflow for Metabolites Annotation on Galalxy via PhenoMeNal named "Metabolomics LCMS/MS XCMS CAMERA, MetFrag processing and annotation(imported from API).

Summary: Keeping all the parameters in the workflow same, every time I run the workflow, I receive different results at the end from Metfrag.

Detail: I have two MS/MS mzml files, one for positive and one for negative mode.
I deploy the instance, open Galaxy and run the Workflow mentioned above. Each tool/step in the workflow runs efficiently and produces same result file (R-file) every time I run the workflow.
The last step of the workflow "metfrag-cli-batch-multiple" causes trouble. So, what happens is, it takes metfrag parameters generated in the last step and the sample metadata, then runs a query to search for any possible matches between the experimental m/z value and its corresponding compound present in wither PUBCHEM or KEGG COMPOUNDS.
At the end, a result csv file is produced, which has every time different number of results.

Now I wrote the procedure in detail to just draw a picture of the issue.
so, If I run this last step three times, I get three different csv results.
I created a Venn diagram to see if they all are similar. So, among three runs, out of 25 m/z values, 18 were common in the three files although all 25 should be common because otherwise it is producing different result everytime, which shouldn't be the case.

Kindly let me know if there is any way to solve this