-
Notifications
You must be signed in to change notification settings - Fork 11
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Expand query graph predicates based on Biolink Model hierarchy #167
Comments
Implemented and merged here: biothings/bte_trapi_query_graph_handler#10 |
It's hard to tell if the predicate expansion is working properly or not, since now the predicate in the query seems to be ignored (#185). This happens even when only 1 predicate is specified. My tentative guess is that the "predicate expansion" work or some other work on the query-making module may have broken the predicate restriction/filtering ability that BTE previously did have. I recall that the "predicate restriction/filtering" used to work - like February, or a month or two ago.
|
Great, let's put this issue on hold pending resolution of #185 |
Now that #185 is closed but this issue was not magically fixed, reassigning this to @marcodarko to look into more. |
This is still not working. For example, the following query should return answers where the edge's predicates are children of the related_to - aka any predicate is fine. Instead, the only answers returned are ones where the edge's predicate is EXACTLY related_to.
|
I think we can use the existing expand predicates functionality of the edge class to include children in the filters, I will try it and see how it goes. |
@colleenXu given the query above, do these results look like what you would expect? Instead I used the expanded predicates of "biolink:related_to" as the post query filters. Sorry about the lengthy response... *EDIT see below for correct query, I made a mistake, the nodes are a little different.
|
@marcodarko I can't tell where this is from, so I'm not sure how to answer. It looks like a Disease MONDO:0004979 -> ChemicalSubstance query, but I just ran that and didn't find those CHEBI IDs. It looks right in that the predicates are descendants of related_to, so I do want those returned. |
@colleenXu my bad, it is Disease MONDO:0004979 > ChemicalSubstance with predicate "correlated_with", sorry! but that's the query I'm running to test this expansion. *EDIT will work further on this, this might not be the results I'm looking for |
@marcodarko That kind of expansion looks correct. However, I would try using some different queries for your testing, because that one seems to depend on clinical risk kp api (which doesn't have positively_correlated_with right now and also has some other issues that I'm dealing with). You could use the query that I gave above (#167 (comment)), or I can give you some other queries that involve predicate expansion... |
This looks resolved now. This query (predicate restriction to "interacts_with" and its children) correctly returns edges with "interacts_with" AND child predicates like "physically_interacts_with" and "directly_interacts_with".
|
In #102, we extended BTE so that it would expand general Biolink node types (e.g.,
biolink:NamedThing
) into all of its subclasses (e.g.,biolink:Disease
,biolink:Gene
). Along similar lines this issue proposes to expand query graph predicates in a similar manner. For example, this query uses the predicatebiolink:causes
(https://biolink.github.io/biolink-model/docs/causes.html), which in turn has antecedants forbiolink:contributes_to
(https://biolink.github.io/biolink-model/docs/contributes_to.html) andbiolink:related_to
(https://biolink.github.io/biolink-model/docs/related_to.html).In the example query above, if I replace the predicate
biolink:causes
withbiolink:related_to
, I expect that the BTE response for the latter will be a superset of the former, but that is not currently the case.The text was updated successfully, but these errors were encountered: