Add tags to datasets

[mfiume]

Add arbitrary text tags that can be used to denote, for example, if the dataset is from case-level observations or curated annotation, whole exome or whole genome, population, geographic region, etc.

[mfiume]

Alternatively, we could add specific fields. I suggest at least the following:

type = { case-level, annotation, any string }
population = { any string }
scope = { genome , exome, any string }

[jrambla]

Are you suggesting a kind of "info" field for the dataset?
In positive +1

[mcupak]

Agreed to postpone to 1.0.

[mfiume]

This is becoming increasingly important as the number of beacons increases.

Also, beacons that serve in silico predictions respond "true" for most alleles. It should be made clear to beacon users that these are not really observed, but hypothetical. We get a lot of excited emails about hits on SolveBio's dbNSFP Beacon for example.

[david4096]

The g4 schemas have a string:List tag that is on the Dataset message for exchanging arbitrary data that is not in named fields.

https://github.com/ga4gh/schemas/blob/master/src/main/proto/ga4gh/metadata.proto#L45

[sdelatorrep]

@mfiume, can the "info" field already present in BeaconDataset be used for this purpose (as @jrambla suggested)?

[mfiume]

Yes, just a matter of whether we'd like to standardize this so that networks can rely on consistent fields. WDYT @sdelatorrep?

[mbaudis]

While in personal use I'm all for "tags", a better way (when having the power to develop in a forward looking way) is to do this in a proper object way. This means:

• no simple "guess what they mean" tags
• no "named attribute for everything"

Similar to the (still not master but pretty much settled) "characteristics" attribute https://github.com/ga4gh/ga4gh-schemas/blob/metadata-integration/src/main/proto/ga4gh/bio_metadata.proto#L101

[sdelatorrep]

@mfiume we believe we should close this issue and open a new one when we have specific tags to include. Or you already have some?

[sdelatorrep]

Any new on this? Should we move the issue to another milestone? Or close it and reopen it when we eventually have some requirements?

[mbaudis]

@sdelatorrep Probably best to do this with concrete implementation. Also, evaluate then use of ontology terms (e.g. EFO) instead of arbitrary tags / typed values.

[susheel]

@mfiume @mbaudis
Suggest we move to next milestone - v1.0
Reason: Requirements for adding tags to datasets are not clear. No real-world use case to test.

[susheel]

Moving to next milestone v0.5