Running abstractor against uniprot with `-m all` errors

[CFGrote]

Hi,
I was trying to get an abstraction for uniprot and ran

abstractor -s https://sparql.uniprot.org/sparql -o uniprot_abstraction.all.ttl -m all -vvvvv

The log reads:

DEBUG:root:Get entities and relation
DEBUG:root:
SELECT DISTINCT ?source_entity ?relation ?target_entity ?mother_source ?mother_target
WHERE {
    # Get entities
    ?instance_of_source a ?source_entity .
    ?instance_of_target a ?target_entity .
    # Relations
    ?instance_of_source ?relation ?instance_of_target .

    OPTIONAL {{
        ?source_entity rdfs:subClassOf ?mother_source .
    }}
    OPTIONAL {{
        ?target_entity rdfs:subClassOf ?mother_target .
    }}
}

After some time (~45 min == uniprot query timeout limit) it finally errors out with

Traceback (most recent call last):
  File "/home/grotec/.conda/envs/default/bin/abstractor", line 118, in <module>
    Abstractor().main()
  File "/home/grotec/.conda/envs/default/bin/abstractor", line 62, in main
    rdf.add_entities_and_relations(sparql.process_query(library.entities_and_relations))
  File "/home/grotec/.conda/envs/default/lib/python3.7/site-packages/libabstractor/SparqlQuery.py", line 182, in process_query
    return self.parse_sparql_results(self.execute_sparql_query(query))
  File "/home/grotec/.conda/envs/default/lib/python3.7/site-packages/libabstractor/SparqlQuery.py", line 99, in execute_sparql_query
    return endpoint.query().convert()
  File "/home/grotec/.conda/envs/default/lib/python3.7/site-packages/SPARQLWrapper/Wrapper.py", line 1107, in query
    return QueryResult(self._query())
  File "/home/grotec/.conda/envs/default/lib/python3.7/site-packages/SPARQLWrapper/Wrapper.py", line 1087, in _query
    raise e
  File "/home/grotec/.conda/envs/default/lib/python3.7/site-packages/SPARQLWrapper/Wrapper.py", line 1073, in _query
    response = urlopener(request)
  File "/home/grotec/.conda/envs/default/lib/python3.7/urllib/request.py", line 222, in urlopen
    return opener.open(url, data, timeout)
  File "/home/grotec/.conda/envs/default/lib/python3.7/urllib/request.py", line 531, in open
    response = meth(req, response)
  File "/home/grotec/.conda/envs/default/lib/python3.7/urllib/request.py", line 641, in http_response
    'http', request, response, code, msg, hdrs)
  File "/home/grotec/.conda/envs/default/lib/python3.7/urllib/request.py", line 569, in error
    return self._call_chain(*args)
  File "/home/grotec/.conda/envs/default/lib/python3.7/urllib/request.py", line 503, in _call_chain
    result = func(*args)
  File "/home/grotec/.conda/envs/default/lib/python3.7/urllib/request.py", line 649, in http_error_default
    raise HTTPError(req.full_url, code, msg, hdrs, fp)
urllib.error.HTTPError: HTTP Error 502: Proxy Error

Is this behaviour to expected? If so, user should be made aware of it. If not, can it be fixed?

Thanks!

[mboudet]

Hello,
Sorry for the delay, for some reason I'm not getting notifications from this repository...
Like you said, it seems like the query is too computation-intensive, and get cut-off by the proxy.
Could you try using the "-m batch" option? It should break the query into smaller subqueries (one for each entity)

Hopefully it should work... I have been meaning to implement LIMIT and OFFSET for these cases, but did not have the time yet.

[CFGrote]

Hi @mboudet, thanks. I tried as you suggested. The command ``` abstractor -s https://sparql.uniprot.org/sparql -o uniprot_abstraction.batch.ttl -m batch -vvvvv ``` produces the following log: ``` DEBUG:root:Get all entities, then, get relations and attributes for each entity DEBUG:root: SELECT DISTINCT ?entity WHERE { ?instance a ?entity . } DEBUG:root:Write RDF (turtle) into uniprot_abstraction.batch.ttl ``` The resulting abstraction file is: ``` $> cat uniprot_abstraction.batch.ttl @Prefix ns1: <http://www.w3.org/ns/prov#> . @Prefix xsd: <http://www.w3.org/2001/XMLSchema#> . [] a ns1:Entity ; ns1:atLocation "https://sparql.uniprot.org/sparql" ; ns1:generatedAtTime "2021-06-22T22:18:50.902685"^^xsd:dateTime ; ns1:wasGeneratedBy <https://github.com/askomics/abstractor> . ```

[mboudet]

So, this is interesting. A far as I can see, this is not an issue with abstractor itself.

It seems this specific request

SELECT DISTINCT ?entity
WHERE {
    ?instance a ?entity .
}

has a weird behavior in uniprot. Instead of returning the results, it return a request, and then return the results in another call.
(You can see it by directly typing the request on the webpage)

If I modify the request (like removing DISTINCT, or adding an empty column on the side), it works.

So, I'm kinda stumped. I contacted uniprot for more information (it could be a bug, it could be an anti-bot feature...)
I could 'fix' it in abstractor by forcing a different request when the target is the uniprot endpoint, bit it's frankly a dirty fix.

I'll wait for an answer from uniprot, and meanwhile see if I can generate an abstraction for you using some local changes.

[mboudet]

I got an answer from the UNIPROT. Since this query is costly, they deal with it differently (redirect to another query)
However, it's currently bugged (no results), since a file is missing..

According to them, it should be fixed in ~ 2 weeks.

[CFGrote]

thanks for your efforts. looking forward to whatever will come back from uniprot.