-
Notifications
You must be signed in to change notification settings - Fork 21
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
Validator is working again and ARAX results are not valid #2046
Comments
I think 3 - 7 will be fixed with the XDTD for KG2.8.0.1c (the current XDTD is still running an old KG2 version). For 1, we should try to pick out something better than I don't understand 8... what's the definition of an abstract class? BiologicalEntity is at depth 2 in the Biolink category tree... |
How about instead of It is not quite the right concept, but maybe close enough? and will quiet the error? |
sounds good to me! I just changed that in |
thanks! @chunyuma can you confirm that you are on a path to solving items 3-7? |
Regarding 8: BiologicalEntity is a abstract: true and I am thinking that this means that there should (must?) not be instances of it. So I suppose the solution is to update KG2 to not use this category any more. |
Steve will create an issue over in KG2 repo |
Yes, the new version of xDTD can solve items 3-7 if KG2c v2.8.0.1 has updated those predicates. |
great, thanks, let's leave this open until we confirm that Which Drugs May Treat Disease X from UI is showing ARAX results that pass all validation, hopefully as soon as we roll out the next xDTD |
Eric where in the code base is the TRAPI validator? |
Yes, we are in discussions with Sierra about whether it can be made |
I think it will not be easy to eliminate use of |
Eric, since Sierra has agreed that the use of RTX/code/ARAX/ResponseCache/response_cache.py Lines 317 to 318 in 7e06641
to catch any validator error corresponding to |
I am thinking that editing the output of the validator dumps() and also the json output would be quite an extensive hack. I think it would be much better to add that exception to the validator itself. Is this only for KG2, or is Sierra's thinking that using BiologicalEntity should be fine for anyone? I think it would be best to start a discussion with Sierra and Richard and ask Sierra if she would approve of the validator being updated to include this exception. What do you think? |
(part of the reason that it would be an extensive hack is that editing validation_messages_text would be one tricky thing, but then line 319 gets the json version of the validator output and then decides on PASS, ERROR, FAIL based on the contents of the json, so that would need to be hacked, too. although that would be easier to hack than the text) |
Here is a snippet of what Sierra told Lili: RTXteam/RTX-KG2#286 (comment)
|
are there any objections to asking Sierra & Richard about adding a formal exception to the validator? |
I don't object. |
@edeutsch understood. Thank you. We will take up the issue with Sierra and Richard. I think there is a good chance we can eventually get an exception in there. In the meantime, please let us know if at some point the errors become sufficiently burdensome/problematic that immediate workarounds (e.g., hacking |
great, thanks. Updating the validator can often be quite rapid when there is a clear path forward. I suggest we wait for that until such a time as it becomes clear that it won't happen quickly. Only then should we contemplate hacking the validator output, I feel. |
The TRAPI validator is on-line again and showing all the TRAPI problems everyone has. We should address these as soon as we can so the validator is showing green for us. Example:
https://arax.ci.transltr.io/?r=140793
"error.knowledge_graph.edge.attribute.type_id.unknown": {
"biolink:original_edge_information": [
I think this means that we are adorning our edges with an attribute "biolink:original_edge_information" but this is not a biolink concept.
Not much we can do about this. I will inquire about this
"error.knowledge_graph.edge.predicate.unknown": {
"biolink:decreases_activity_of": [
I don't think this is a valid predicate any more?
"biolink:directly_interacts_with": [
{
"edge_id": "NCBIGene:2353--biolink:directly_interacts_with->UniProtKB:P0DPQ6"
}
],
],
"biolink:increases_activity_of": [
"error.knowledge_graph.node.categories.not_concrete": {
"FMA:82762": [
{
"categories": "['biolink:BiologicalEntity']"
I think we're not supposed to use abstract classes?
Can we make a plan to address these 8 items?
The text was updated successfully, but these errors were encountered: