search

cdanielmachado / smetana

SMETANA: a tool to analyse interactions in microbial communities
Other
56 stars 11 forks source link

Cross-feeding interaction untraceable metabolites #29

Closed iuliachiciudean closed 2 years ago

iuliachiciudean commented 2 years ago

Hi Daniel,

I generated my Smetana microbial community interactions, but I can not seem to find a logic in the exchanged metabolites. Is there any way to trace a metabolite production pathway in the donor organism?

I am verifying the BlastKoala-generated pathway of my donor organism, and I am looking at the metabolic model using Fluxer (https://fluxer.umbc.edu/).

For example, from the Smetana metabolic interactions, it appears that my donor organism is donating H2S, but I can not find a pathway for H2S production in the genome. By visualising the GEM with Fluxer I can see that H2S is needed (not produced) in the assimilatory sulfate reduction pathway to produce L-Cysteine.

So how is Smetana deducing that the metabolite is produced by the microorganism? Any suggestions on how I can trace the Smetana production of the metabolites in the genome or GEM?

Thank you! Best regards, Iulia

cdanielmachado commented 2 years ago

Hi Iulia,

I would recommend just running FVA on the model to check if it really can or cannot secrete H2S.

This is what SMETANA does for the community model to calculate the MPS (metabolite production score) of the donor species for each compound.

iuliachiciudean commented 2 years ago

Hi Daniel,

This is the FVA result from CNApy for 2 metabolites that are donated according to Smetana (scour 1). H2S: H2S Methanol: Methanol

Is this what you suggested to look for? Those results are not what I have expected for an organism that is donating those compounds for community growth.

Do you think something went wrong during the Smetana calculations? I can send you the metabolic model .XML file.

Best regards,

Iulia

cdanielmachado commented 2 years ago

It seems that your model can secrete H2S, so the output from smetana is correct. If that is not the expected behavior you might need to manually curate your model.