Add assay_type as an additional, more precise, experiment descriptor

[dbujold]

The JSON data hub already includes a proposed "assay_type" identifier, but this should be formalized in the metadata spec, in order to clearly define what goes in there. "assay_type" gives more fine-grained details about the nature of the experiment. For example, if experiment_type is DNA Methylation, then assay_type might be Bisulfite-Seq or 450K or others. What goes in here should be formalized to avoid ("WGB-Seq" vs "Bisulfite-Seq" vs "BS-Seq" vs etc.).

So, the proposal is to have:

• "experiment_type" as a controlled vocabulary following the currently metadata specification: ["DNA Methylation", "Histone H3K4me1", "ChIP-Seq Input", "mRNA-Seq", "Chromatin Accessibility", ...]
• "assay_type" that will repeat the label for the chosen ontology term in EXPERIMENT_ONTOLOGY_URI.

assay_type examples: For "http://purl.obolibrary.org/obo/OBI_0001863", asssay_type = "shotgun bisulfite-seq assay" For "http://purl.obolibrary.org/obo/OBI_0002004", assay_type = "Methylation 450K BeadChip"

[dbujold]

From Sita:

I agree with the proposal. However, note that people have been submitting data for a while with the old spec, so tooling around spec will have to be not so strict when dealing with this (especially as the added attributes are computable from attributes in the old spec). In general, I would be against adding such new computable attributes (as it adds no new information), but I think the tag name ASSAY_TYPE vs EXPERIMENT_TYPE is informative.

I propose that we have a class of attributes that should ideally be in the spec, but if absent would be computed in some documented/standardized way from the required attributes.

[dbujold]

Added non-mandatory field "assay_type" to experiment_attributes.