Unexpected prediction of eukaryotic RIG-I-like receptor signaling pathway from microbial 16S data

[AlirezaDostmohammadi]

Dears,

I hope you are doing well.

I am using PICRUSt2 for functional prediction analysis based on my 16S rRNA sequencing data. After predicting the abundance of functions, I moved to the KEGG pathway level to infer the abundance of pathways, based on the predicted function profiles.

I then performed differential abundance analysis to identify pathways that are significantly different between my groups. One of the pathways that showed significant differences is:

ko04622 – "RIG-I-like receptor signaling pathway"

As far as I understand, this pathway is specific to eukaryotic host cells, particularly innate immune cells in animals, and is not something that microbes themselves express.

So:

Does the database used in PICRUSt2 account for the microbiome's impact on host (human) pathways?

Or is this prediction likely a false positive that should be ignored, given that this pathway is not expressed by microbes?

Thank you in advance for your clarification.

[R-Wright-1]

Hi there,

I actually wouldn't recommend using PICRUSt2 for KEGG pathways unless you have a subscription to KEGG and can replace the pathway file with one of your own. We don't have these pathways output as default for this reason, and you can see a little more information on this here. I think if you are finding this in microbes, it's likely that it's just because KEGG doesn't differentiate between the two, just says which KOs contribute to different pathways. We have previously filtered the MetaCyc pathways to only microbial ones (not in the current version), so you could do something similar to this if you wanted to.

Best wishes,
Robyn

[AlirezaDostmohammadi]

Dear Robyn,

Thank you for getting back to me.

Is the MetaCyc pathways database used in PICRUSt2 the latest version, or do I need to update it manually?

Could you please tell me how to update the MetaCyc or KEGG pathways database in PICRUSt2?

Best,
Alireza

[AlirezaDostmohammadi]

@R-Wright-1 Dear Robyn,

I hope you're doing well.

I'm writing to follow up on my earlier message regarding the MetaCyc and KEGG pathway databases used in PICRUSt2. I was wondering whether the MetaCyc database included by default is up to date?

Could you please advise on how to update the MetaCyc or KEGG pathway databases in PICRUSt2?

I’d really appreciate any guidance you can provide.

[R-Wright-1]

Hi there,

Sorry for the delay. I've been away either teaching workshops or preparing to teach them so haven't had a chance to get to these questions. To update the pathway databases in PICRUSt2 you would need to create equivalent files to the default ones included with PICRUSt2. I'll give some details on where they have come from.

I took the MetaCyc pathways information from what is provided with HUMAnN3 (version 24). I have not ever looked into updating this personally, but what you would need is a file that maps from EC numbers to MetaCyc reactions (you can see the current one in picrust2/default_files/pathway_mapfiles/ec_level4_to_metacyc_rxn_new.tsv) and then a second file that is a structured mapping file that contains information for mapping reactions to MetaCyc pathways (again, you can see the current one in picrust2/default_files/pathway_mapfiles/metacyc_pathways_structured_filtered_v24.txt).

For KEGG, you would need to replace the two files picrust2/default_files/pathway_mapfiles/KEGG_modules_to_KO.tsv and picrust2/default_files/pathway_mapfiles/KEGG_pathways_to_KO.tsv. I don't know where these came from as they were from the last publicly available KEGG release. I don't have a subscription to the FTP so can't say exactly, but it looks like you could get these from here.

Robyn