Open schristley opened 2 years ago
@bzimonja comments - seems like this would be simple, any gotchas?
@schristley would you want to do this just for the count endpoints? As you know I like symmetry but doing this on the other endpoints would result in a lot of data...
Should we send only results that have counts, or grab all repertoire ids first, and report the ones that have 0 counts?
@schristley would you want to do this just for the count endpoints? As you know I like symmetry but doing this on the other endpoints would result in a lot of data...
I don't have a use case where I'd want all that data in one shot. I think VDJServer's usage will be very similar to the Gateway's in that it will be shown on request by the user, and I'm comfortable with lazy-loading the data as needed. I haven't written any code to utilize those endpoints yet though.
Should we send only results that have counts, or grab all repertoire ids first, and report the ones that have 0 counts?
I think I'd prefer only results that have counts, similar to facets. I guess it's okay to assume a 0 count means no stats are available if that's somehow easier...
For me, it's easier to only pull stuff that has counts, since I have them in a collection separate from repertoires, so anything with no stats shouldn't be in it :) But not that difficult to add "count > 0" to the query or post-processing.
Just like the
repertoire
endpoint in the ADC will return all the repertoires with an empty query, can we do the similar thing for thecount
endpoints in the stats API? I think maybe that can be faster than sending hundreds (or thousands in VDJServer's case) of repertoire_ids in the request and parsing/querying that on the server side?VDJServer's use case is to show those counts, so it's more efficient to issue one query to get all of them, versus issuing hundreds/thousands of individual queries.