PMID:33134189 PA2146 Gene Knockout Is Associated With Pseudomonas aeruginosa Pathogenicity in Macrophage and Host Immune Response

[CuzickA]

This is a publication that @JonMWilkes has selected to curate.

Curation link https://canto.phi-base.org/curs/c60773877297f4b1/ro/

[JonMWilkes]

The 'autofill' on selecting Q9I1W9 from uniprotkb returns the name Q9I1W9_PSEAE, while the gene is called PA2146, both in the paper and on the UniprotKB page. Is it (a) possible to define the gene name when the Uniprot ID is first entered and/or (b) can this be propagated through the annotations already made for this gene?

[CuzickA]

Checking this session

Fig 1

edit to

@JonMWilkes, when you are ready, please go ahead and contact UniProt to update the gene name as discussed.

There is a mechanism to edit the genotype name from the 'Pathogen genotype management page'

Select 'edit details'

Select 'edit' and then enter the required name in 'allele name'

Select 'finish'

This is a temporary fix while waiting for the UniProt name to be updated which can sometimes take a while.

[ValWood]

Hemorrhage seems a useful clinical phenotype. However, there are presumably existing ontologies to use for these phenotype (HPO or MP)?. As you move to annotate a lot of human pathogen interactions you should plug in a suitable existing ontology and contribute to that ontology instead of putting them in PHIPO (the situation is a bit different than for pathogen and pathogen-host phenotypes where no preexisting resource existed)and share the data with other resources (like Monarch)

[CuzickA]

This sounds like a good idea. How would we use these other ontology terms in PHI-Canto? Would they need to be created as PHIPO terms with a cross-ref to the other ontology?

[ValWood]

No, you would just use it as an ontology instead of PHIPO.
You need to decide the scope of the mammalian phenotypes that you want to curate, and find the correct ontology.
The ontologies presented could be different for different species , so if somebody adds a human identifier, instead of PHIPO you would use an appropriate ontology like HPO for single species annotations. The important thing is not to extend your ontology unnecessarily to cover terms that are the scope of existing obo foundry ontologies.

[CuzickA]

Is it possible in PHI-Canto to load more than one ontology / namespace into the PHI and single species phenotype annotation workflows?

[JonMWilkes]

I am starting another A.fumigatus paper in an mammalian infection model and human granulocyte killing assay. Have we made any decisions on the feasibility/desirability of utilising other ontologies for coding and classifying responses and results?

[CuzickA]

Is it possible in PHI-Canto to load more than one ontology / namespace into the PHI and single species phenotype annotation workflows?

@jseager7 do you know if this is possible in PHI-Canto?

[CuzickA]

I am starting another A.fumigatus paper in an mammalian infection model and human granulocyte killing assay. Have we made any decisions on the feasibility/desirability of utilising other ontologies for coding and classifying responses and results?

For now, I suggest using the terms we have in PHIPO and capturing any terms we currently don't have in the 'new term selection' option.

[jseager7]

Is it possible in PHI-Canto to load more than one ontology / namespace into the PHI and single species phenotype annotation workflows?

@CuzickA Not that I'm aware of. Canto's configuration suggests that annotation types can only be linked to a single OBO namespace. Without code changes, I think we only have two options:

1. Add a new annotation type for human or mammalian phenotypes and link it to the relevant ontology, or

2. Get all the ontologies we need and patch them so that every term has the namespace we need (e.g. single_species_phenotype or pathogen_host_interaction_phenotype.

Option 2 is technically less work, but it's a pretty bad hack since we'd be forced to overwrite the namespace properties of the original terms (OBO terms can only have one namespace property). So it also won't work if we need to make the phenotypes available to both single-species and pathogen-host interaction branches, since we'd need more than one namespace per term for that.

Option 1 is more flexible, and should be able to make the phenotypes available for both single-species and multi-species annotations, but it would mean more workflows for the curators to follow. Also, if these phenotypes need all the same annotation extension configuration as existing phenotypes (pathogen phenotypes, host phenotypes, or pathogen-host interaction phenotypes) then I'd need to duplicate the configuration, which adds more maintenance work (though the extension configuration syntax might provide ways to share some configuration rules).

The 'ideal' solution for this problem would be to allow annotation types to link to multiple OBO namespaces, but I have no idea how much work that would involve.

[jseager7]

Although, now that I think about it, if we're considering bringing in terms from ontologies that are completely agnostic about pathogens and hosts, why do we need the single-species branch in PHIPO at all?

If we do need to make some distinction between pathogen and host phenotypes even in the single-species branch, then it could be ambiguous to bring in terms that were never designed with that distinction in mind. I'm a bit wary of having the PHI-Canto annotation type name being the only semantic indication of whether the phenotype was expressed by a pathogen or a host.

If we don't need the distinction, then it implies that PHIPO could be reduced to only the pathogen-host interaction phenotypes and that the single-species phenotypes could be cobbled together from other phenotype ontologies (though this would probably require multiple namespaces per annotation type).

[ValWood]

I don't think the single-species branch should really be in PHIPO. I think this was done to make ontology editing easier (and was a bit of a step gap). Ideally, you would have single species phenotypes to use the most relevant ontology depending on the species. For the fungal pathogens there's no ontology available. There was talk of checking status of the available plant ontologies, but I think they were not being actively developed when we loked.

[JonMWilkes]

@CuzickA - I have sent the above correction request to Uniprot - I think they have changed the format since you put together the document you sent me - hope this works.

[CuzickA]

Thanks @JonMWilkes

The confirmation email from UniProt has arrived in the curation inbox. I will move it to the UniProt correspondence folder.

Here is the ticket number

[CuzickA]

I found these in the Mammalian Phenotype Ontology

My thoughts are

1. that we do not want the organ specific level term as we have already captured 'lung' in the AE.

2. the term 'Hemorrhage' on its own does not give us what we want. We would need to create terms like these

If we keep the above annotation we will need terms for 'hemorrhage'
WT 'host hemorrhage with pathogen present'
mutant 'increased host hemorrhage with pathogen'

3. Does it seem appropriate to make these new PHIPO terms and use the 'MP' term 'Hemorrhage' as part of the logical definition?

[ValWood]

That probably does make sense.

[CuzickA]

Fig 2b

I'm trying to figure out the 'key' bit of information here. Should this expt be annotated as

1. 'presence of pathogen growth within host' based on the intracellular bacterial counts in Fig 2b
2. or are we looking at a host immune response with the host cells phagocytosing the bacteria? Do we need to add new terms for phagocytosis?
   

The results section heading is 'PA2146 Deletion Has No Impact on RAW264.7 Phagocytosis'
and the intro text in this section is 'Phagocytic cells play a central role in eliciting responses to
acute P. aeruginosa infection. Phagocytosis can kill bacterial cells and present antigens to other immune cells (Pryjma et al.,).'
The text relating to Fig 2b is 'Though, the deletion of PA2146 showed moderate tendency of inhibition
activity against RAW264.7 phagocytosis-mediated killing (Figure 2B, Left panel), there was no statistical significance
between wild type of PAO1 and PAO11PA2146 at 2 to 4 h treatment (Figure 2B, Right panel).'

Which probably favours option 2 above

any thoughts on this @JonMWilkes, @ValWood?

[CuzickA]

Once queries above for Fig 2a and 2b are resolved this session will be complete.

[CuzickA]

Looking at my query above for Fig 2a, I have made comments in PHI-base/phipo#366 and created NTR: abnormal host morphology during phagocytosis of pathogen in PHI-base/phipo#370

[CuzickA]

For Fig 2b
NTR: host immune system phagocytosis of pathogen present

Once NTRs loaded this session can be approved

[CuzickA]

NTRs added to session. Session approved. Closing ticket.

[CuzickA]

Gene name now updated in UniProt https://www.uniprot.org/uniprotkb/Q9I1W9/entry

Needs updating in PHI-Canto.

[jseager7]

@CuzickA The gene name is updated in the session now:

It seems Canto needed to be restarted after the gene cache was refreshed in order to update the gene name in the user interface.

However, Canto is showing the entry name (RGVIR_PSEAE) instead of the gene name (PA2146, technically the name of the ordered locus). This is probably just a missing case in Canto's code, so I'll open an issue on the Canto tracker for that.