bio-ontology-research-group / aberowlweb Goto Github PK

I've finally figured out what was bothering me about the API routes! I think there's a faulty rewrite rule (although it could be a problem of my browser, lets see if you can reproduce it). Steps to repeat:

1. browse to http://www.aber-owl.net/api/dlquery/?type=subeq&labels=true&query=hypertension&offset=1
2. Correctly, resource not found
3. Browse to http://www.aber-owl.net/api/dlquery?type=subeq&labels=true&query=hypertension&offset=1
4. Note that the URL is rewritten to that in step 1, and is not found

This only happens when 'www' is in the URL, so I guess it's a problem of the reverse proxy automatically adding a / on the end when that is the case?

Example: protein from SIO ontology

Index PMC and MEDLINE and perform queries directly from DL Queries; add a new tab for this purpose.

it's a pretty small issue, but can be confusing, e.g.

http://www.aber-owl.net/api/dlquery/?labels=true&query=%27has%20part%27%20%20some%20(%27increased%20pressure%27%20and%20(%27inheres%20in%27%20some%20(blood)))&type=subeq&offset=1&ontology=HP

then removing the / to make it

http://www.aber-owl.net/api/dlquery?labels=true&query=%27has%20part%27%20%20some%20(%27increased%20pressure%27%20and%20(%27inheres%20in%27%20some%20(blood)))&type=subeq&offset=1&ontology=HP

makes it work

There should be a way to change the status to running as soon as ontology is synced or reloaded without having to restart ontology server.

See OBOFoundry/OBOFoundry.github.io#1053
The OBO prefixes are not correctly (or rather: not consistently) used. Import from OBO should be fixed.

e.g. from http://www.aber-owl.net/ontology/HP/#/Browse/%3Chttp%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FHP_0001409%3E

clicking around on the left tree view seems to send two dlqueries for every class clicked, one of which returns 301. Here's a screenshot of the network tab from clicking around on HP:

Enable annotation of text provided by user; select ontologies; select method; output annotations.

Add examples (autofill on click) to the SPARQL query tab. E.g., use the examples or templates from the paper.

It would be better to make searchclasses case insensitive

Discriptive Logic should be "Description Logic" throughout.

The cursor on the frontpage should be within the query field, so that I can just start typing.

Trying to query for
'is attribute of' some entity on the front page will fail; should return at least classes from SIO.

OWL queries should separate between subclass, superclass, equivalent class, queries, and allow (by checkbox) to include all results or only direct sub-/super-classes.

At least keep the log of complex queries, and add them to a file; make the file available for download (describe in Help page).

Add to API:

• autocomplete
• ElasticSearch
• SPARQL query rewriting
• DL Queries
• Literature search (when implemented)

examples:

http://www.aber-owl.net/api/dlquery?labels=true&query=%27part%20of%27%20some%20%27apoptotic%20process%27&type=subeq&offset=1

http://www.aber-owl.net/api/dlquery?labels=true&query=%27has%20part%27%20%20some%20(%27increased%20pressure%27%20and%20(%27inheres%20in%27%20some%20(blood)))&type=subeq&offset=1

weirdly, simple queries of a single concept are working: http://www.aber-owl.net/api/dlquery?labels=true&query=hypertension&type=subeq

steps to repeat:

browse to http://www.aber-owl.net/ontology/ENM/#/Browse/%3Chttp%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FBFO_0000001%3E

note that the definition of the annotation property itself, and other properties of the annotation property, also seem to appear as an annotation property

e.g. in this case 'A phrase describing how a class name should be used. May also include other kinds of examples that facilitate immediate understanding of a class semantics, such as widely known prototypical subclasses or instances of the class. Although essential for high level terms, examples for low level terms (e.g., Affymetrix HU133 array) are not' which I guess is the definition of example of usage,

and 'PERSON:Daniel Schober' who is an editor, among others.

I also suspect that properties defined for this class in other ontologies are shown

Search results are not returned in the order returned by ElasticSearch (which uses boosting to prioritize certain fields and prioritizes exact matches over partial matches); for search of names and ontology names, ElasticSearch ordering should be used.

Create a link to the Github Issue tracker on the main AberOWL website (in the footer).

the tree view does not load, see: http://www.aber-owl.net/ontology/PhenomeNET/

It looks like there's some kind of error: from the API: http://www.aber-owl.net/api/dlquery/?type=subeq&labels=true&query=hypertension&ontology=PhenomeNET

Server side pagination might help better the performance of search results

I see that the querynames API has changed to _find. Could I pelase have a list of annotation properties that are included in the 'synonyms' key for that model? Maybe it would be good to put this in the documentation, also

This is the case for even simple queries like:

SELECT DISTINCT ?p where {?s ?p ?o.}

and

prefix owl: <http://www.w3.org/2002/07/owl#> 

SELECT distinct ?ontology
WHERE {
    ?ontology a owl:Ontology.
}

Steps to repeat:

Compare http://aber-owl.net/ontology/CMO/#/ with https://bioportal.bioontology.org/ontologies/CMO/?p=summary

Due to the way classes are being indexed/searched, there are many classes that show up in the Bioportal search, that do not show up in AberOWL. I think that:

• Intermediate words are not searched, e.g. try typing 'creatine,' which should ideally come up with the 'urine creatine [...] set of terms
• Synonyms are not searched, e.g. try searching 'thigh girth,' which should come up as a synonym of 'proximal hind limb circumference'

The first issue particularly can make it seem like lots of classes are missing from the ontology

The file seems to contain passwords and local config files for our AberOWL instance. Could they be removed (and passwords changed), and could the repository be made public?

According to the SPARQL protocol, response format should be determined by content negotiation [1]. Most SPARQL clients use this (including RDFlib), which makes the AberOWL endpoint inaccessible through most libraries.

[1] https://www.w3.org/TR/sparql11-protocol/#conneg

There should be a checkbox which allows the retrieval of ALL (not only direct) sub- or superclasses based on a DL query.

Searching for "protein" on the main page does not give the result from the SIO ontology.

Falsey values seem to count as true values. So the following will return fine:

http://aber-owl.net/api/dlquery?type=equivalent&query=%3Chttp://purl.obolibrary.org/obo/HP_0000822%3E

While this:

http://aber-owl.net/api/dlquery?type=equivalent&query=%3Chttp://purl.obolibrary.org/obo/HP_0000822%3E&labels=false

will return nothing. It would also be nice if you could pass falsey values to e.g. ontology to run without that flag. This would make the client code for API interaction make more sense/ a bit nicer

Show instances in AberOWL, and allow DL Queries to retrieve instances

Add example queries on the front page. Use one searchable by label, one by OBO ID, one by ontology, one by DL query. Auto-change the tabs according to the query.

The issue is probably because the index setting are not the same on elasticsearch server new site.

querynames seems to have been partially replaced by searchclasses. now, querynames doesn't return any results if given an "ontology" parameter. since they both have the same type of result, could these endpoints be consolidated? if not, perhaps an error message could be added to the querynames endpoint, if calling it with an ontology parameter, that guides you to the searchclasses endpoint.

There are no axioms shown on the class overview within the ontology browse view; e.g., [http://10.254.145.17/ontology/SIO/#/Browse/<http%3A%2F%2Fsemanticscience.org%2Fresource%2FSIO_000000>] should show axioms in which "entity" is involved (subclass, equivalent class, disjoint), but none are displayed.

The idea is to look for a classes starting with a keyword in specified list of ontologies. For example if the user try to find phenotype classes in both HP and MP starting with a keyword they should be able to have classes from both ontologies.

e.g. http://aber-owl.net/api/dlquery?type=subeq&labels=false&query=%3Chttp://purl.obolibrary.org/obo/HP_0005549%3E&offset=1&ontology=HP&deprecated=true

I think it's a useful feature if you're using the AO API to look up data annotations, since you can use the annotation properties of the deprecated class to find out what it has been replaced with

The AberOWL API needs to be documented and examples created, and there should be a link to the API on the main page.

e.g. http://aber-owl.net/api/dlquery/?labels=true&type=equivalent&direct=false&query=hypertension&ontology=HP&axioms=true

It would be great if you could pass html=false or something and get the class axioms stripped of their HTML (just in plain Manchester OWL Syntax)

The weights of the matches in different fields need to be updated.
Currently, searching by OBO ID gives no results. IDs should be higher than Labels should be higher than Synonyms should be higher than Definition should be higher than rest. This needs to be implemented in ElasticSearch indexes.

Add a LICENSE.md file to Github describing the AberOWL license.

It seems not possible to get the Falls class from MEDLINEPLUS via dlquery using its label, although it's available via its IRI

compare: http://www.aber-owl.net/api/dlquery/?type=equivalent&query=falls&offset=1&ontology=MEDLINEPLUS&labels=true

with

http://www.aber-owl.net/api/dlquery/?type=subeq&query=%3Chttp%3A%2F%2Fpurl.bioontology.org%2Fontology%2FMEDLINEPLUS%2FC0085639%3E&offset=1&ontology=MEDLINEPLUS

Steps to repeat:

1. Browse to http://www.aber-owl.net/ontology/ENM/
2. Expand 'entity' in the left tree
3. wait a long time
4. Every class in the ontology will open under entity

If you click on one, changing the current window url to http://www.aber-owl.net/ontology/ENM/#/Browse/%3Chttp%3A%2F%2Fsemanticscience.org%2Fresource%2FCHEMINF_000049%3E then reload the page, the hierarchy will load correctly.

My guess would be that direct=false, for some reason, but the tools show that the query run is:
http://www.aber-owl.net/api/dlquery/?type=subclass&direct=true&axioms=true&query=%3Chttp%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FBFO_0000001%3E&ontology=ENM

Could this parameter be failing for some reason? I suppose the reason it works upon reload is getPathToRoot doesn't subclass query entity. Could there be some eccentricity of the ontology itself causing everything to be a direct subclass of entity?

Ability to pass e.g. inferences=false to dlquery to receive only results asserted directly, without inferences made by the OWL reasoner.

I'm not sure how often this feature would be useful, but for my case it is because I am trying to use AberOWL to text mine axioms. Consider the following set of classes: http://www.aber-owl.net/ontology/ENM/#/Browse/%3Chttp%3A%2F%2Fpurl.bioontology.org%2Fontology%2Fnpo%23NPO_284%3E

I want to use the subclasses of 'two-dimensional nanoparticle' as possible entities for my axioms, so I can form e.g. "nanotube and has_component_part some carbon". However, in the case of carbon nanotube, this axiom already exists. Therefore, because of the reasoner, 'carbon nanotube' is a subclass of nanotube. It is not desirable to include these inferences as possible entities for axioms.

This could be achieved with the OWLAPI structural reasoner, but it might be a bit of a hit on resources since I guess you'd have to maintain the 'classified' ontology in memory. Maybe something smart could be done by comparing the results of the reasoner query with the structure of the explicit ontology?

it would be a powerful feature for aberowl if you were able, via the API, to access entities referred to in the database cross references described by classes, such as to SNOMED, ICD-X, UMLS, etc.

for 'databases' that are ontologies in aberowl, it should be quite easy. medical terminologies harder: as I understand it, one of the major problems is that we don't have any licence agreement to 'make subsets of them available programmatically', or whatever the licence says.

so, how about literally just redirecting the API call to Bioportal? e.g. http://data.bioontology.org/ontologies/SNOMEDCT/classes/http%3A%2F%2Fpurl.bioontology.org%2Fontology%2FSNOMEDCT%2F410607006

at the least I think it would make a nice follow-up to the synonym expansion paper... but it would be quite a useful resource, particularly for text mining and such, too.

bioportal has mappings: https://bioportal.bioontology.org/mappings/show/DOID?target=http://data.bioontology.org/ontologies/HP but i wonder if limiting them to the ones ontologies define themselves would help accuracy etc

@reality reports:
AberOWL's queryNames API endpoint is returning nothing

@aliraza995 could you check what's happening please?

right now it doesn't seem to make it in there, which can be a problem for text mining applications

when I do a DL Query in AberOWL and click on any of the results, I never get taken to the actual description of the class; it just goes back to an empty page.

Enables access to AberOWL through SSL