Should our response to VLM API requests include additional fields on top of those required by the spec?

[jennifer-bowser]

The "official" VLM API spec outlines specific fields that it expects to be provided by responses to requests. The protocol only allows for very high-level data to be returned - outside of some metadata about the responding node and the data sources it contains, this official response mainly consists of a list of:

1. A simple count of the number of variants of a given Zygosity that match the requested variant
2. The ID of the source for this data

This means that other data that may be helpful is not included, such as:

• The VRS ID that AnyVLM normalized the queried variant to
• A list of the variants that AnyVLM found as a match for the query
• The list of CohortAlleleFrequencyStudyResults that AnyVLM found for the query

This left us (the devs) with two questions:

1. Should we provide any extra information in our response to VLM API requests in addition to the required fields?
2. If so, what extra information should we provide?

[jennifer-bowser]

In my opinion, if we do decide that our responses should include additional information, we should tackle that after our MVP release. I worry about adding anything else to our plate with the deadline so close.

I do think that the data we're returning is really bare-bones, so I'd definitely love to include some extra information. But I think that since the VLM API spec is in such an early stage, we might have a chance to influence its development - if we agree on what additional info we'd like to include, it'd be worth reaching out to the VLM API folks to see if they'd be willing to incorporate the additions we'd like to add into the official spec. I don't know how receptive they'd be to change requests, though.

[jsstevenson]

agree that this is probably the lowest priority -- although as long as nothing requires new data, I think this would be the equivalent of a 1- or 2-point ticket

[jsstevenson]

My vote -- we should also provide the data formatted as a VA-Spec CAF (which would include the VRS allele as the subjectVariant)

• adhering to standards is good
• putting out data that adheres to OUR standards is even better