Discuss existing case deposition APIs and come up with shared spec

[jxchong]

as discussed in the Baltimore meeting, GeneMatcher, PhenomeCentral, and Decipher all have case deposition APIs and MyGene2 is planning to create one. If we can agree upon a spec for what a common MME case deposition API would look like (no one has to commit to actually implementing it now), MyGene2 won't have to reinvent the wheel and create yet a 4th incompatible standard.

Perhaps start with describing each of the existing nodes' APIs?

[harindra-a]

matchbox is a pure bare metal server and everything that goes in, out and matches off it, is via an API. It has an authenticated,

HTTPS POST /patient/add

endpoint where a patient (as per MME spec) can be inserted into the system to be matched. Currently it is hidden behind a firewall and only the /match endpoint is exposed to the world via a proxy for extra security.

Our seqr web-app acts as the UI for this and constructs the patient object to be inserted from data aggregated in seqr when the analysts pushes a button (and fine tunes the submission if needed).

So theoretically, IF we go this route, it is possible to simply expose that /patient/add endpoint to the world so anybody bona fide can insert a patient. The caveat as of now is, it expects the MME patient schema.

[fschiettecatte]

The GeneMatcher API is described here.

Also we distribute a python client implementation, which is used in PhenoDB so we know it works 😀

[buske]

The API for depositing cases into PhenomeCentral is described here. It's currently implemented by PhenoTips and Patient Archive.

[Relequestual]

THe DECIPHER API is still under development, but we are seeking feedback from a number of institiutions including Gemics Englaland. As such, I'll email you all the link as opposed to posting it publically for now.