Mimosa no_spec_param error

[Jalapenobadger]

Hey Cecilia,

I have had one issue. Trying to run the data through the online version of Mimosa2 I get the following error: Error: object 'stoichReac' not found get_species_cmp_scores species network !rxn_param rxn_param F no_spec_param humann2_param rel_abund_param

I went through and removed all metabolites without KEGG IDs, and I've tried renorming by rel abundance using the GalaxyTrakr-Humann2 tool. I've attached the data I'm trying to run if that helps.

On Tue, Jun 30, 2020 at 12:46 PM Cecilia Noecker notifications@github.com wrote:

Hi Pete, Thanks for your interest in MIMOSA2! This is an interesting question. You are correct that MIMOSA2 is more focused on data interpretation and hypothesis generation (i.e. what microbial mechanisms can explain metabolite observations), while Melonnpan is more focused on metabolite prediction. For each metabolite, MIMOSA2 fits a simple linear model relating microbial metabolic potential to metabolite measurements. So it is possible to take the MIMOSA2 model results from one dataset and apply them to another dataset to predict metabolites, but I would only expect that to be informative if the two datasets are from very similar sample populations. Any resulting predictions should definitely be considered with a grain of salt.

We also address this question a little bit in the MIMOSA2 FAQ here: https://borenstein-lab.github.io/MIMOSA2shiny/faqs.html#cmpsAlone

Yes, MIMOSA2 will work with as few as 12 samples. You'll be less likely to find significantly predicted metabolites than you would be with a larger sample size, but you'll still be able to see which metabolites are explained better or worse by shifts in microbiome metabolic potential.

Hope that helps!

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/borenstein-lab/MIMOSA2shiny/issues/4#issuecomment-651914403, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACQB6LMJ4DECAO73K5I6TATRZIJGXANCNFSM4OLM6O5Q .

Originally posted by @Jalapenobadger in https://github.com/borenstein-lab/MIMOSA2shiny/issues/4#issuecomment-652507574

[Jalapenobadger]

I attached the data in the reply email to the last thread, here it is again in xlsx format:

Metab_Kegg.xlsx

MergedSequenceFeatures.xlsx

[cnoecker]

Hi, Thanks for sharing your data files. Your metabolite file looks great, but unfortunately, your taxa file won't work as provided. MIMOSA2 only knows how to link microbiome data to reactions and metabolites via a limited number of input formats. The table has to provide either: 1) 16S rRNA ASV sequence variants, in a table providing either the sequences themselves or SILVA or Greengenes OTU IDs, not just taxonomic classifications. or 2) Providing functional information directly in the form of KO abundances. (These can be stratified abundances with quantities for each KO in each taxon, as is provided in the output of Humann2.) Humann2 has a utility function to map Uniref families to KOs if you have metagenomic data that you've analyzed with that option.

You can see example tables for each option here: https://borenstein-lab.github.io/MIMOSA2shiny/input.html# Let me know if that makes sense and works for your data. Feel free to also suggest any clarifications that would be helpful in the input or docs.

[Jalapenobadger]

Yes, that all makes perfect sense, and it sounds like I should be able to get it to work still. I was given the data from Humann2, so I will try to find the Humann2 utility; hopefully it is included in the GalaxyTrakr Humann2 suite. I'm sure the docs make sense, I am so new to all this a lot of the info just goes over my head. I think I was just looking at the example input for the taxa and assumed it would operate agnostic to the format of the taxa names, because the K # 's looked to me like an arbitrary assignation rather than needing to conform to anything. Thanks again! -Pete

On Thu, Jul 2, 2020 at 1:42 PM Cecilia Noecker notifications@github.com wrote:

Hi, Thanks for sharing your data files. Your metabolite file looks great, but unfortunately, your taxa file won't work as provided. MIMOSA2 only knows how to link microbiome data to reactions and metabolites via a limited number of input formats. The table has to provide either: 1) 16S rRNA ASV sequence variants, in a table providing either the sequences themselves or SILVA or Greengenes OTU IDs, not just taxonomic classifications. or 2) Providing functional information directly in the form of KO abundances. (These can be stratified abundances with quantities for each KO in each taxon, as is provided in the output of Humann2.) Humann2 has a utility function to map Uniref families to KOs if you have metagenomic data that you've analyzed with that option.

You can see example tables for each option here: https://borenstein-lab.github.io/MIMOSA2shiny/input.html# Let me know if that makes sense and works for your data. Feel free to also suggest any clarifications that would be helpful in the input or docs.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/borenstein-lab/MIMOSA2shiny/issues/5#issuecomment-653139171, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACQB6LODF673I5LS7VGWKYTRZTBHPANCNFSM4OPAQ6RQ .

[cnoecker]

Hi Pete, Glad to hear it, thanks for your input. In case you haven't found it, the Humann2 function is humann2_regroup_table, as described here: https://bitbucket.org/biobakery/biobakery/wiki/humann2#rst-header-regrouping-genes-to-other-functional-categories I'm not sure if it would be in the GalaxyTrakr suite or not.

[Jalapenobadger]

There is a function simply called regroup, but based on trying to run it today I actually don't think it has the same functionality, so I'm going to look into what you've just linked this weekend. I would have started with this but I'm a little daunted by the idea of getting the proper Kegg database loaded locally since I understand it is no longer free or some sort of subscription. Thanks for the heads up. -Pete

On Fri, Jul 3, 2020 at 9:11 PM Cecilia Noecker notifications@github.com wrote:

Hi Pete, Glad to hear it, thanks for your input. In case you haven't found it, the Humann2 function is humann2_regroup_table, as described here: https://bitbucket.org/biobakery/biobakery/wiki/humann2#rst-header-regrouping-genes-to-other-functional-categories I'm not sure if it would be in the GalaxyTrakr suite or not.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/borenstein-lab/MIMOSA2shiny/issues/5#issuecomment-653704044, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACQB6LN5WROSIEX7GFDKINDRZZ6TNANCNFSM4OPAQ6RQ .

[cnoecker]

Hi Pete - Humann2 is able to map the genes to KEGG Orthologs without actually needing to get access to the KEGG database (which does unfortunately require a subscription). I think I shared the wrong link before - this one describes more specifically how to first download the mapping files and then run the regroup_table function. Hope that works for you!

[Jalapenobadger]

Ah, gotcha. I'll try that out.

On Fri, Jul 3, 2020 at 11:04 PM Cecilia Noecker notifications@github.com wrote:

Hi Pete - Humann2 is able to map the genes to KEGG Orthologs without actually needing to get access to the KEGG database (which does unfortunately require a subscription). I think I shared the wrong link before - this one https://bitbucket.org/biobakery/humann/wiki/Home#markdown-header-humann2_regroup_table describes more specifically how to first download the mapping files and then run the regroup_table function. Hope that works for you!

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/borenstein-lab/MIMOSA2shiny/issues/5#issuecomment-653713245, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACQB6LORM2DFT7OIKBIV6LLRZ2L37ANCNFSM4OPAQ6RQ .