Usefulness of having "experiment_type” and “assay_type” in two separate fields

[dbujold]

Feedback received from Sveinung Gundersen during the Open for Comments period:

7. Another point of feedback that just came up when trying to adopt the Experiments metadata attributes into our harmonised data model: https://docs.google.com/spreadsheets/d/1wqlcnYHwT9yQVCj525n41LQJENbgHeQfn7V572e8ev8/edit?gid=1471635995
   What is the reasoning behind including both “experiment_type” and “assay_type”. It seems experiment_type (e.g. “DNA sequencing assay”) is only a generalization of assay_type (e.g. “whole genome sequencing assay”), but such generalizations should (at least in principle) be possible to derive from the ontology (in the example here, “DNA sequencing assay” is the parent term of “whole genome sequencing assay”. Also, if the point is to specify the molecule that are sequenced, there is a “molecule_type” field exactly for that!

I am filling out example data for our data model, and in the chosen example, which is a “ChIP-seq assay”, I have a hard time determining what should go into “experiment_type”. Should that be the more general “ChIP assay”, the even more general “DNA sequencing assay”, or something else?

I am now leaning towards thinking that the “experiment_type” field is unnecessary.

[dbujold]

Thank you for this @sveinugu, I will close this issue now since #5 is closed.

[sveinugu]

@dbujold Well, with my comment in #5, I tried to convey that I am not convinced by your argument:

However, we felt we should maintain the separation for the cases where a controlled vocabulary / free text is used, to enable the easy separations at these two levels.

I still think experiment_type is unnecessary together with molecule_type and assay_type and will probably not include it for that reason in the FAIRification of Genomic Annotations adoption.

First, I don't think use cases that are not providing ontology terms should be encouraged to the level that the design caters to such use.

And second, based on the example values of experiment_type I don't see a scenario where there is a need to separate between RNA sequencing assays and DNA sequencing assays as such, where this cannot be relatively easily solved by the molecule_type field?

If this is just to retain compatibility with INSCD, perhaps this can be maintained in a way that does not add unnecessary confusion to users?

I am re-opening the issue to await further comments to my concerns.

Edit: actually, I don't have the permissions to re-open...

[sveinugu]

Also, the new "How to fill" description of experiment_type does not really help the user to select e.g. between "ChIP Assay" and "DNA Sequencing Assay" (from the "ChIP-seq assay" example I highlighted in my initial feedback):

Specify the broad experiment type, as the assay type property will provide the finer granularity.

One would need to consider the example values to get an idea of what the field is meant to cover. This is perhaps an indication that the overall meaning of the field itself is not clear enough to retain in the standard?

With now (at least) three independent comments on the overlap between the three fields, perhaps the choice to retain them all should be reconsidered: #11 ?