apiEntityMap needs to be updated for genotype

[yuanzhou]

I'm working on genotype expansion for phenogrid. Once all the matches returned from the http://beta.monarchinitiative.org/compare API, I'll need to add them into the target data list, and each genotype will also need a type attribute to be specified.

Right now we are using the apiEntityMap's apifragment property as the desired type. Some genotypes have the same 'MGI' prefix, what should I do to add the genotype type? Otherewise I can't distinguish genotype from gene. @harryhoch

And some genotypes have the MONARCH prefix instead of MGI.

apiEntityMap: [ 
                {prefix: "HP", apifragment: "disease"},
                {prefix: "OMIM", apifragment: "disease"}, 
                {prefix: "ZFIN", apifragment: "gene"},          
                {prefix: "MGI", apifragment: "gene"}
            ]

[harryhoch]

@yuanzhou, Good questions. I'm not sure that I know the answers. @kshefchek ?

[kshefchek]

Typically it's not possible to determine ID type based on prefix (HP excepted in that list). I could add type to any json object, where is this needed?

[yuanzhou]

This is used in the dataLoader: https://github.com/monarch-initiative/phenogrid/blob/master/js/dataloader.js#L167

I think I could get around this by not using this apiEntityMap but manually add genotype as the type of each genotype. Aside from this, the apiEntityMap is only used by the dataLoader to determine the type of other targets. And this type info is also used in the tooltip of grid cell.

Maybe I should get rid of this apiEntityMap from phenogrid, but this assumes that mouse and fish are only use gene, human uses disease. Any suggestions?

[kshefchek]

Is the issue that owlsim does not return type information?

[yuanzhou]

There are id and label in every return now. It will be great if we add type there too, I'm not sure if every id/label really has a unique type. E.g., MGI can be used for either gene or genotype. If we add the type info along with the ID, at least it resolves this issue.

[kshefchek]

Let me see if this is something I can address on the Monarch API level.

[yuanzhou]

Thanks @kshefchek please keep me posted on this. If it can not be addressed on the monarch api level, I'll try to find a workaround and get rid the use of this.

[yuanzhou]

@harryhoch, not sure how long I'll wait until @kshefchek has this issue addressed in the API level. Can I assume the type of targets under human is always 'disease', and type of targets under fish/mouse is always 'gene' or 'genotype'? Because right now the genotypes are being added after all the species targets are loaded. If my assumption is true, I can simply specify the type based on the species name instead of the prefix. And I can specify the type of all the added genotypes as 'genotype' later, since only fish/mouse has this feature for now.

[harryhoch]

@yuanzhou, I think this is a fine approach for now.

[nlwashington]

FYI, i hesitate to add any assumptions to logic within phenogrid. it should not know what a "disease" is.

phenogrid should be generic enough to render anything from human with the HP prefix. some things that are annotated with HP are diseases (as is what you are getting from owlsim at the moment), but in the future it could be anything: genes, genotypes, variants, people. please, please don't hardcode anything!

[nlwashington]

furthermore, i do think that you ought to be given the types in some way; whether that is from some kind of monarch-api call, or from owlsim directly, you need the info. at the moment, owlsim doesn't store anything about types and it'll be some time until that's refactored/added; but since you are getting the genotypes from monarch app, the schema could include a "type" : genotype for example, together with the id and label. if you have a preference for how this looks, please specify.

[harryhoch]

@yuanzhou, @nlwashington makes an excellent point here. Please sketch out what you would need to make this code sufficiently generic and flexible. Let's not hardcode anything that we can parameterize...

[kshefchek]

I think this will be fixed with my PR or am I misunderstanding? https://github.com/monarch-initiative/monarch-app/pull/1031

[yuanzhou]

@kshefchek I was aware of your PR fix. Will make the changes once your PR get merged. Thanks!

[yuanzhou]

@kshefchek after your PR https://github.com/monarch-initiative/monarch-app/pull/1031 got merged, I could see the type in the compare API returned JOSN. But the type is still null from the initial simsearch POST.

http://beta.monarchinitiative.org/simsearch/phenotype

with the following POST data:

input_items=HP:0000726+HP:0000746+HP:0001300+HP:0002367+HP:0000012+HP:0000716+HP:0000739+HP:0001332+HP:0001347+HP:0002063+HP:0002067+HP:0002172+HP:0002322+HP:0007159+HP:0009466+HP:0001972+HP:0003502+HP:0005190+HP:0001763+HP:0003318+HP:0001371+HP:0001380+HP:0001394+HP:0001156+HP:0001159+HP:0001004+HP:0010886+HP:0006530+HP:0004792+HP:0002816+HP:0000954+HP:0000974+HP:0000765+HP:0000602+HP:0002240+HP:0030084+HP:0000049+HP:0000028+HP:0000175+HP:0000204+HP:0000202+HP:0000316+HP:0000343+HP:0000349+HP:0002023+HP:0000327+HP:0002055+HP:0000394+HP:0000463+HP:0000431+HP:0000494+HP:0000484+HP:0000486+HP:0000508+HP:0009748+HP:0007018+HP:0001773+HP:0001769+HP:0003311+HP:0001544+HP:0001508+HP:0003196+HP:0001249+HP:0001187+HP:0001169+HP:0012774+HP:0002650&target_species=7955

[kshefchek]

Made a small tweak and cleared the cache, it should be working when beta updates.

[yuanzhou]

@kshefchek Tested again and it worked! Thanks a lot!