review of assays with detection techniques

[hectorguzor]

We recently specified the definition of a detection technique as an assay (A) that is part of another assay (B), where the target entities of A and B are disjoint, and the target entity of A is a proxy for the target of assay B.

James created a SPARQL query that returns a TSV file with assays that have a detection technique. The TSV file has four columns. Column A includes assays that have a detection technique, column B includes these assays’ target entities, column C includes their detection technique and column D includes the detection technique’s target entities.

I collected all assays that have target entities that are not disjoint with the target entities of their detection techniques, or where it was not clear to me that there was disjointness because the target entities were not specified. You can fine those assays Here. In light of changes to the detection technique columns in the assays template, it would be good to review these assays to ensure the modeling is correct. I started by reviewing 3 assays. These are highlighted in yellow. Column E includes some notes on what I take to be the issue with the modeling and column F has notes on how we can go about correcting the modeling.

For example, in row 4 FAIRE-seq has DNA sequencing assay as a detection technique. Currently, 'DNA sequencing assay' and 'FAIRE-seq assay' have region as their target entity. This violates the disjointness condition. In this case I recommend we import regulatory_region from SO and make it the target entity of 'FAIRE-seq'. We should also make primary structure of DNA macromolecule the target entity of 'DNA sequencing assay'. If we import 'regulatory_region' and make it a sibling of 'primary structure of DNA macromolecule' then we can explain disjointness.

[DanBerrios]

Discussed on OBI Dev Call 06/13/22:

• “Regulatory region” should be imported from SO
• “Region” conflates nucleotide and AA sequencing
• Some of the assays on the spreadsheet should have “primary structure of DNA” as target for sequencing assay
• Probably we need to do some remodeling of “primary structure of DNA”, moving out from under region
• Also relevant: COB modeling and SO, see https://docs.google.com/presentation/d/1UhWNyZI6_4tc-xR2NXNAIaqBTDgOYLYCAG7emrkGmKY/edit

[jamesaoverton]

In general, if we see a problem with a detection technique using this SPARQL query, the right solution might be to

1. change the parent
2. change the detection technique
3. remodel more deeply
4. rethink the SPARQL query and/or our new rule

For example, the query might be telling us that what we put as a detection technique should really be the parent assay.

For FAIRE specifically, people on the call today were proposing moving "primary structure of DNA macromolecule" and "primary structure of RNA" out from under "region", but I'm not sure where else they should go.

[hectorguzor]

Discussed in OBI call 2022-08-15: I will work on a PR to address issues with ‘FAIRE-seq’. I will import regulatory_region from SO and make it the target entity of 'FAIRE-seq'. I will also make primary structure of DNA macromolecule the target entity of 'DNA sequencing assay'. 'regulatory_region' will be a sibling of 'primary structure of DNA macromolecule'. This will help explain the disjointness of the two target entities.

We still need to discuss whether 'primary structure of DNA macromolecule' should be a subclass of 'region', or be placed somewhere else.