biological associations are reversed

[tmcelrath]

@tmcelrath : I think this was fixed, right Dmitry? @proceps : I am not sure. I remember, that there was some issues in the original FM database. Some of the relationships were flipped there. And there was no easy way to separate which are right, which are wrong. The best way, probably is to try to download the table with all relationships, and see how those are resolved.

[tmcelrath]

Okay: Mating with- changed to "mating with" and made reflexive and transitive.

To fix:

• [x] Attendance [deprecated] - flip all and replace with "Localized to". See details below.
• [x] is Predator of - added RO definition, keep
• [x] is Prey of - added RO definition, keep
• [X] Parasitism [deprecated] - destroyed, replaced by "Attached to"
• [x] Host plant [deprecated] - 12k records, flip all and replaced with "Localized to". ~250 records at the "start" are correct. Done - Host plant destroyed after transfer.
• [x] Pollination [deprecated] - 1949 records, flip all and replace with "Localized to".
• [x] Reared [deprecated] - destroyed, replaced by origin relationships in TW
• [x] Host fungus [deprecated] - destroyed, replaced by "localized to"

[tmcelrath]

@proceps @mjy1 - see annotations. I think Attendance, Host Plant, and Pollination can all be easily fixed programmatically right now.

[tmcelrath]

@proceps Looking at the data.

• For Attendance, the following associations are present in the DB: Hemiptera-Hymenoptera, Hymenoptera-Hemiptera, Coleoptera-Plant The original intention of the relationship is Hemiptera-Hymenoptera, Hemipterans producing mildew and are being visited by ands and bees to feed on it. Majority of the records are Hemiptera-Hymenoptera, few are inverted. In 3i database I renamed it to mutualistic relationship, and keep the relationship Hemiptera-Hymenoptera. Do you really want to change all of those? My understanding, that only non hemipterans should be moved out of the list?
• Host plants. We have plant-insect and insect-plant associations, My understanding you want make all of those insect-plant and change type to "Localized to"?
• pollination is plant-insect association, you want to make it insect-plant and changed to "localized to"

[tmcelrath]

RE Attendance: 1) I see 351 records in INHS-IC database for Attendance. 2) I do not see the "Coleoptera-Plant" ones. Can you provide a link? 3) We are transitioning (on suggestions of @mjy1) away from explicit assumptions of "positive-negative" biological interactions. Labels generally do not make these assumptions, and only demonstrate physical interactions (e.g. something on a flower might not be pollinating, but just visiting it for nectar). These suggestions are a movement towards that. 4) There is no OBO Relationship Ontology for "Attendance" - I would suggest maybe "trophically interacts with": http://purl.obolibrary.org/obo/RO_0002438 or "co-localizes with": http://purl.obolibrary.org/obo/RO_0002325. 5) Hymenoptera are the "attendants" and Hemiptera are "Attended by", correct? Anything that is the inverse of this should be flipped (most of the records I think). I think the correct ones are more recently edited?

RE Host plants/Pollination: 1) All SHOULD have collection object as SUBJECT. We don't track "this plant Collection Object was visited by this insect OTU" in the database. We should ONLY have "this insect Collection Object was localized to this plant OTU". Otherwise no data would be reported/exported via DWC Export. 2) Hence, both "Pollination" and "Host-plants" should be corrected to "Localized to" and the insect collection object should be the SUBJECT, never the OBJECT.

@mjy1 correct?

[tmcelrath]

@proceps For being more specific about relationships please use OBO relationship ontology as a reference https://www.ebi.ac.uk/ols4/ontologies/ro

[tmcelrath]

All non-deprecated Biological Relationships now have OBO RO definitions in definition:

[tmcelrath]

Should "localized to" be changed to "adjacent to" in order to fit better with RO? http://purl.obolibrary.org/obo/RO_0002220

[tmcelrath]

@proceps Looking to attendance, could not find coleoptera. But still there is a problem. CO is always object, but in some cases it is Hemiptera, in some other cases it is Hymenoptera. How would you divide those?

[tmcelrath]

For "Attendance (=adjacent to)" - there are two groups currently. Both are reversed in different ways.

1. Current: Hemiptera (subject) OTU "attendant to" Hymenoptera (object) Collection Object
   • Fix: Flip to Hymenoptera (subject) Collection Object "adjacent to" Hemiptera (object) OTU
   • Set in question: https://sfg.taxonworks.org/tasks/biological_associations/filter?per=500&biological_relationship_id%5B%5D=15&extend%5B%5D=object&extend%5B%5D=subject&extend%5B%5D=biological_relationship&extend%5B%5D=taxonomy&extend%5B%5D=biological_relationship_types&page=1&descendants=true&object_taxon_name_id%5B%5D=581784&taxon_name_id_mode=true
2. Current: Hymenoptera (subject) OTU "attendant to" Hemiptera (object) Collection Object
   • Fix: honestly? I see no evidence of this coming from a collection object, and there are <100 of them, so I think they can be deleted.
   • Set in question: https://sfg.taxonworks.org/tasks/biological_associations/filter?per=500&biological_relationship_id%5B%5D=15&extend%5B%5D=object&extend%5B%5D=subject&extend%5B%5D=biological_relationship&extend%5B%5D=taxonomy&extend%5B%5D=biological_relationship_types&page=1&descendants=true&subject_taxon_name_id%5B%5D=581784&taxon_name_id_mode=true

[tmcelrath]

@proceps from my experience both are valid relationships. Since the majority of the relationships coming from Hymenoptera side, I would suggest to keep one type of the relationship and always put hymenoptera (CO or OTU) on the subject side.

[tmcelrath]

@proceps There is visited_by relationship

[tmcelrath]

I like "visited by" but it lacks a definition. I'm having trouble coming up with a definition that can be explicitly differentiated from "adjacent to" without including assumptions of directional benefits or costs. My opinion is honestly to just fold into "localized to/adjacent to" here: http://purl.obolibrary.org/obo/RO_0002220 @mjy1 opinions?

[tmcelrath]

@proceps mutualistically interacts with? http://purl.obolibrary.org/obo/RO_0002442

[tmcelrath]

[tmcelrath]

[tmcelrath]

@mjy For the purposes of curation of an insect collection I would ensure that the subject is the collection object, this just makes curation easier if consistent. The subject should further be the one with the CatalogNumber if there are 2 COs and one doesn't have a CatalogNumber. If there are two COs both with CatalogNumber then it will be come tricker which to make the subject in reflexive relationships (like localized_to), I'd suggest some arbitrary decision, like alphabetical position of taxon determination name decides subject and object.

Regarding which relationships to use I think using generic ones like co-located, and localized to are going to be fine. One thing to remember, not sure how it impacts things, we're about to (well, this year sometime) introduce AnatomicalPart, and "Flower" can be an antomical part, so we can do this:

CO -> AnatomicalPart -> Some OTU

bee -> some flower -> some flowering plant

or

mite -> a thorax -> thorax of some CO

My only thinking is along the lines of whether we lose our ability to refactor to this granularity if we switch off of polinator now. For example we might see that labels include "flower" for some relationships that use polinator.

[tmcelrath]

[tmcelrath]

Manually fixed by @tmcelrath transferred by @mjy reversed by @proceps - closed.