I’m running HUMAnN 3.7 on human stool data and am currently trying to make sense of some pathways results annotated by MetaCyc. Indeed, HUMAnN identifies several pathways annotated as “C4 photosynthetic carbon assimilation cycle” (ex. PWY-241, PWY-7115, PWY-7117) which surprises me quite a bit since these pathways are plant specific and MetaPhlAn only identifies bacteria in my samples. These pathways are never connected to any taxon (in the hierarchical HUMAnN outputs) so I guess they were picked up based on diamond annotations.
A collegue suggested it could be leftovers of plant DNA coming from food intake, would this even be possible? Would you have a suggestion on how I could investigate these results further to understand them better?
I’ve had this thought too, and it seems possible, but it could also very well be that a microbial pathway has genes that look similar enough that they are annotated as photosynthetic. For example, this bug is a cyanobacterium, but lives in the gut…
One thing you could check is whether humann thinks it knows what bug they pathway came from.
To build on @kescobo’s comment - HUMAnN uses a gap-filling procedure that allows exactly one reaction to be missing from a pathway and still call the pathway as present. This can sometimes result in a pathway X being called because it’s very similar (shares a lot of reactions) with another pathway Y. That usually happens when you have subtle variants of a pathway in different domains of life though. I would think that the photosynthetic pathways would be quite distinct from non-photosynthetic pathways, meaning that these are more likely to be true detections.
Agreeing with your logic @franzosa , turning off gap-filling does indeed not change the results much.
Indeed @kescobo all of these pathways - and other “surprizing” ones such as engineered pathways - are mostly found at community level, so unfortunately no further explanation there…
I think my best bet is to manually filter these pathways (=engineered or photosynthetic) out before downstream analysis then, unless you see a reason not to? I’d be curious to know how others are dealing with them!
