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] Please don't change current REST services

Ray Fergerson ray.fergerson at stanford.edu
Thu Dec 19 11:54:58 PST 2013


Eamonn,



We don’t currently have the notion of a score for the annotations. It
wasn’t terribly obvious that the previous scores were of much general
value so we didn’t regenerate them for the new API. Because of the way
that the annotation is done, we don’t actually have information about the
“quality” of the match itself. The system just does exact string matches.
If you are interested in scoring based on prefName vs. synonym then you
can easily construct such a scheme yourself since we provide the data.



Ray



From: Paul R Alexander [mailto:palexander at stanford.edu]
Sent: Thursday, December 19, 2013 10:51 AM
To: Eamonn Maguire
Cc: Ray Fergerson; ISA Team; support at bioontology.org
Subject: Re: [bioontology-support] Please don't change current REST
services



I’ll let Ray handle the question about the score because I’m not entirely
sure what the plans are there.



We are working on ways to make more of the class information available
when annotating without suffering performance degradation. Hopefully this
will make it into a release early in 2014.



Paul





On Dec 19, 2013, at 3:06 AM, Eamonn Maguire <eamonnmag at gmail.com> wrote:



Some notes on the annotator api:



Score is missing from results now. SYN (synonym) and PREF (preferred) is
the only thing available. Is score now gone from your comparison or are
you just not showing it?



'prefLabel' would be useful in the annotated class information otherwise
no way of knowing the actual ontology term that can be used for annotation
without additional queries. I have to do an additional query for each
result Item just to find out it’s string representation.



Thanks as always,



Eamonn



On 19 Dec 2013, at 09:28, Eamonn Maguire <eamonnmag at gmail.com> wrote:



Great, thanks Paul.



1. I’m assuming here that you mean the links and context still appear. You
can turn these off by adding &no_links=true&no_context=true. We didn’t
document these options because we weren’t sure if people would need them,
but it does help in slimming down large responses quite a bit.



This is great. The one thing we’d need that appears in the links section
is the Ontology the term comes from. Would it be possible for you to move
this to a higher level so that the whole links section can be skipped if
we’re just doing a term search?



I’m working on the ontology tree traversal code now so I’ll probably have
more questions by the end of the day.



Thanks,



Eamonn





On 18 Dec 2013, at 18:03, Paul R Alexander <palexander at stanford.edu>
wrote:



I think we could examine raising the page size limit. We would need some
time to evaluate the ramifications, though. I’ve added an issue for us to
look into this.



The paging is not done by slicing the results, at least not in code. It’s
done at the database level, so the underlying system there may do some
optimizations when subsequent requests that differ only in the offset are
made. We could examine doing something like you suggest with pre-fetching
subsequent pages.



Paul





On Dec 17, 2013, at 2:56 PM, Eamonn Maguire <eamonnmag at gmail.com> wrote:



Well the problem is the speed (500 take a long time to load) and not being
able to make it completely parallel since I'll need the first call to know
how many pages to retrieve.

Also, how are you implementing the paging? Are you doing the query each
time then slicing the results according to the page size and requested
page number (to create an offset)? Or are results cached in some way to
enable pre-emptive speed ups of subsequent calls?

In summary, one call would be better for us, especially if implementation
is as I described in the former.

  _____

From:  <mailto:ray.fergerson at stanford.edu> Ray Fergerson
Sent: 17/12/2013 22:46
To:  <mailto:eamonnmag at gmail.com> Eamonn Maguire
Cc:  <mailto:support at bioontology.org> support at bioontology.org;
<mailto:isatools at googlegroups.com> ISA Team
Subject: RE: [bioontology-support] Please don't change current REST
services

Eamonn,



You can just ask for the next page. If you have 1000 items and the page
size is 500 this is just one extra call. Which call specifically is giving
you problems?



The way to specify pages is at the top of the documentation at:



 <http://data.bioontology.org/documentation>
http://data.bioontology.org/documentation



Ray



From: Eamonn Maguire [mailto: <mailto:eamonnmag at gmail.com>
eamonnmag at gmail.com]
Sent: Tuesday, December 17, 2013 8:31 AM
To: Ray Fergerson
Cc:  <mailto:support at bioontology.org> support at bioontology.org; ISA Team
Subject: Re: [bioontology-support] Please don't change current REST
services



Hi Ray,



We're working on porting all our code over to the new version of bioportal
(I can refer you to the Java code if you’d like it for other users so they
don’t have to do it themselves).



I have a question. I may have a few more by the time I finish. One is to
do with paging. How do I remove paging. There is a limit to the calls of
500 items per page. However, I have some examples, e.g. dose where the
number of results is quite large, nearing 1000. Is there a way of skipping
the paging altogether?



Thanks,



Eamonn



On 7 Nov 2013, at 00:10, Ray Fergerson <
<mailto:ray.fergerson at stanford.edu> ray.fergerson at stanford.edu> wrote:



Eamonn,



Sorry for the delay in getting back to you. The idea here would be to
group all of the preferred name lookups into a single call and retrieve
them all at once using the “batch” call. Please see the documentation for
how to do this.



 <http://data.bioontology.org/documentation#nav_batch>
http://data.bioontology.org/documentation#nav_batch



We may (probably will) later have a scheme for specifying which properties
of a class you want to retrieve on all calls. This feature only exists on
certain calls at the moment, including the batch call.



Ray



From:  <mailto:bioontology-support-bounces at lists.stanford.edu>
bioontology-support-bounces at lists.stanford.edu [
<mailto:bioontology-support-bounces at lists.stanford.edu>
mailto:bioontology-support-bounces at lists.stanford.edu] On Behalf Of Eamonn
Maguire
Sent: Friday, October 25, 2013 4:28 AM
To:  <mailto:support at bioontology.org> support at bioontology.org Support
Cc:  <mailto:isatools at googlegroups.com> isatools at googlegroups.com
Subject: [bioontology-support] Please don't change current REST services



Hi all,



First of all, we use bioportal REST services across numerous tools,
including OntoMaton and ISAcreator, ISAconfigurator etc. This move to the
SPARQL results really messes things up for whilst providing an extra
number of calls to get the same information we get with one call at
present. For example:



This call to search on melanoma gives me this.



{

·         "page": 1,

·         "pageCount": 45,

·         "prevPage": null,

·         "nextPage": 2,

·         -

"links": {

o    "nextPage":  <http://data.bioontology.org/search?q=melanoma&page=2>
"http://data.bioontology.org/search?q=melanoma&page=2",

o    "prevPage": null

},

·         -

"collection": [

o    -

{

§  "prefLabel": "Melanoma",

§  "@id":  <http://purl.bioontology.org/ontology/MDR/10053571>
"http://purl.bioontology.org/ontology/MDR/10053571",

§  "@type":  <http://www.w3.org/2002/07/owl#Class>
"http://www.w3.org/2002/07/owl#Class",

§  -

"links": {

§  "self":
<http://data.bioontology.org/ontologies/MEDDRA/classes/http%3A%2F%2Fpurl.b
ioontology.org%2Fontology%2FMDR%2F10053571>
"http://data.bioontology.org/ontologies/MEDDRA/classes/http%3A%2F%2Fpurl.b
ioontology.org%2Fontology%2FMDR%2F10053571",

§  "ontology":  <http://data.bioontology.org/ontologies/MEDDRA>
"http://data.bioontology.org/ontologies/MEDDRA”,

§  
.



Now, if I want to find out which ontology this term came from, I have to
do another search to this  <http://data.bioontology.org/ontologies/MEDDRA>
http://data.bioontology.org/ontologies/MEDDRA



I have to perform 2 calls to just get the name of the ontology it came
from. It requires 2 calls for each term and places extra burden on both
your servers and slows down our applications.



Is there some way you could you make it easier to get this information
from the original call please?



Many thanks,



Eamonn



--

Eamonn Maguire





_______________________________________________
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/20131219/0f40bef0/attachment-0001.html>


More information about the bioontology-support mailing list