Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[bioontology-support] Populating ontologies on NCBO virtual appliance, and filtering annotator results

Paul R Alexander palexander at stanford.edu
Fri Mar 22 15:56:16 PDT 2013


Colin,

Just to confirm, you are running the script with these settings:
$FROM = "http://rest.bioontology.org/bioportal"
$TO = "http://{your domain or ip}:8080/bioportal"

And I assume you are running it directly from the Appliance? I just ran the script to import all ontologies from BioPortal and it worked reasonably well. There were some initial errors due to running out of disk space, but once I resized the VM it seemed to work ok.

Paul


On Mar 22, 2013, at 10:35 AM, Colin Morris <colin at sciencescape.org> wrote:

> I just tried it and got the same error message.
> 
> 
> On Thu, Mar 21, 2013 at 8:47 PM, Paul R Alexander <palexander at stanford.edu> wrote:
> Colin,
> 
> I think this was due to some data problems on the NCBO REST services. Can you give it another shot?
> 
> Thanks,
> Paul
> 
> 
> On Mar 20, 2013, at 3:01 PM, Colin Morris <colin at sciencescape.org> wrote:
> 
>> Thanks so much for the fast response. I tried running the import script and got this traceback:
>> 
>> ./bioportal/ontology_wrapper.rb:149:in `[]': can't convert String into Integer (TypeError)
>>         from ./bioportal/ontology_wrapper.rb:149:in `initialize'
>>         from ./bioportal/ontology_wrapper.rb:148:in `each'
>>         from ./bioportal/ontology_wrapper.rb:148:in `initialize'
>>         from ./bioportal/BioPortalRestfulCore.rb:1702:in `new'
>>         from ./bioportal/BioPortalRestfulCore.rb:1702:in `generic_parse'
>>         from ./bioportal/BioPortalRestfulCore.rb:1701:in `each'
>>         from ./bioportal/BioPortalRestfulCore.rb:1701:in `generic_parse'
>>         from ./bioportal/BioPortalRestfulCore.rb:487:in `getOntologyList'
>>         from /usr/lib64/ruby/gems/1.8/gems/activesupport-2.3.14/lib/active_support/core_ext/benchmark.rb:17:in `ms'
>>         from /usr/lib/ruby/1.8/benchmark.rb:308:in `realtime'
>>         from /usr/lib64/ruby/gems/1.8/gems/activesupport-2.3.14/lib/active_support/core_ext/benchmark.rb:17:in `ms'
>>         from ./bioportal/BioPortalRestfulCore.rb:487:in `getOntologyList'
>>         from ./import_ontologies.rb:51:in `import_ontologies'
>>         from ./import_ontologies.rb:105
>> 
>> I set $API_KEY_TO (using the API key associated with the admin account on my local instance) and $API_KEY_FROM, and left everything else as it was. I ran the script directly (./import_ontologies.rb) rather than the way the FAQ recommended, since that resulted in it being interpreted as a shell script (import_ontologies.rb: line 4: =: command not found, etc.)
>> 
>> I'm afraid I've exhausted my limited Ruby debugging capabilities (the error seems to relate to user, which is a global variable defined outside of ontology_wrapper.rb?).
>> 
>> 
>> On Wed, Mar 20, 2013 at 2:06 PM, Paul R Alexander <palexander at stanford.edu> wrote:
>> Colin,
>> 
>> Thanks for your interest. We don't include the ontologies because, amongst other reasons, many people who use the Appliance are interested in something other than the biomedical domain. I'm curious if you've already tried the web services and found them slow?
>> 
>> There is a script that is included on the Appliance that can facilitate importing ontologies directly from BioPortal. It won't work for every ontology for a few reasons:
>> 1) Some ontologies have licensing restrictions that prevent their distribution
>> 2) Some require special processing to work correctly with the system's back-end
>> 
>> Here's instructions on using the import script:
>> http://www.bioontology.org/wiki/index.php/Virtual_Appliance_FAQ#How_can_I_migrate_ontologies_from_BioPortal_or_previous_NCBO_Virtual_Appliance_versions_into_a_new_Appliance.3F
>> 
>> Each RRF ontology in BioPortal is actually multiple RRF files which are submitted as a single zip file (make sure there are no subdirectories). I've CCed Lian, who works with the system that handles RRF ontologies. Lian, do RRF files from UMLS need any special processing?
>> 
>> I'll defer to someone more knowledgeable on the Annotator CUI question, though I believe you would need to use the concept id and do a lookup for that concept on the BioPortal REST services to see the CUI.
>> 
>> Let me know if you have further questions or run into problems.
>> 
>> Paul
>> 
>> 
>> On Mar 20, 2013, at 9:48 AM, Colin Morris <colin at sciencescape.org> wrote:
>> 
>>> Hi,
>>> 
>>> We're interested in using the NCBO annotator in an application of ours. Because we have millions of short documents that we'd like to annotate, we thought we'd run the virtual appliance locally (to speed things up on our end, and avoid imposing on your servers too much). 
>>> 
>>> I was (maybe naively) expecting that the image would include the same ontologies as on bioportal.bioontology.org. Is there an easy way to reproduce something like that environment locally? That is, easier than locating each of the ~335 ontologies currently on the site and adding them through the web interface? (We'd like to cast as broad a net as possible)
>>> 
>>> Or, given a copy of the most recent release of the UMLS (which we have), is there an easy way to plug it into our local NCBO instance? (I see there's a way to add RRF files as ontologies, but there are 42 associated with our UMLS install, and I suspect that just dropping them all in won't have the desired effect.)
>>> 
>>> 
>>> And a mostly unrelated question: Given the results of a call to the Annotator web service, is there any way to get the UMLS CUI of each returned concept? I see that the concept element has an 'id' and a 'localconceptid', but neither of those seem to correspond to UMLS. 
>>> 
>>> I see that the annotator has the option to expand its results to infer new results that have the same CUI as found concepts. What I'd like to do is the opposite -- get distinct concepts, by CUI.
>>> 
>>> Any help very much appreciated!
>>> 
>>> Thanks,
>>> Colin
>>> _______________________________________________
>>> bioontology-support mailing list
>>> bioontology-support at lists.stanford.edu
>>> https://mailman.stanford.edu/mailman/listinfo/bioontology-support
>> 
>> 
> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/bioontology-support/attachments/20130322/5942452a/attachment.html>


More information about the bioontology-support mailing list