Re: [Psidev-ms-dev] [Psidev-ms-vocab] Release candidate 3.74.0_rc1 of psi-ms.obo

[Daniel S. <dsc...@ip...>]

Dear all.

I tend agree with Andy here. For DL ontologies where the is_a needs to 
be logically correct at all times for automated reasoning, the OBO 
Foundry naming conventions won't allow errant classifications like
/ no/ /threshold is_a statistical threshold/
But for PSI, where we have a simple taxonomic CV driving an intuitive 
user friendly parameter annotation, I  guess its fine in the pragmatic 
sense. You could read the terms
as being linguistic ellipses, where the implicit postfix " /indicating 
parameter/" is left out, e.g. a better but longer name would be
/no/ /threshold indicating  parameter is_a statistical threshold/ 
indicating /parameter/

How this would be represented in OWL DL is dependent on the type of 
reasoning you envision, which is impacted by closed vs. open world 
assumption / the expressivity regime of your choice. That is why I 
always fought for defining best-practice patterns for a particular 
regime (use case in mind), i.e. as an ontology design pattern in a 
corresponding pattern repository, e.g.
http://ontologydesignpatterns.org/wiki/Submissions:*Parameter*

In formal ontologies one could allow such indicators under an 
appropriate information artefact/parameter/threshold 
parameter/parameter_indicating_no_threashold
in case of a simple taxonomy, or for the case of axiomatic class 
definitions,  one of the many possibilities using BFO and a lacks object 
property is mentioned in
http://ontology.buffalo.edu/medo/negative_findings.pdf
But all this is a bit remote for our scope ;-)

Best regards,
     Daniel.

Am 09.03.2015 um 12:14 schrieb Jones, Andy:
> Hi Steffen,
>
> We need to put this term into data formats so it is clear that the user is saying - I did not use a threshold on my data. If there is a blank, we cannot distinguish between a badly formatted file (i.e. I used a threshold but forgot to tell you) or the no threshold case. The PSI-MS CV is generally a pragmatic collection of terms rather than an ontology, but I agree it is sensible for us to be aware of good practice in ontologies. Under good practice ontology guidelines, how is it expected that one can distinguish the "null" case from the "zero" case - as in this example?
>
> Best wishes
> Andy
>
>
>
>
> -----Original Message-----
> From: Steffen Neumann [mailto:sne...@ip...]
> Sent: 09 March 2015 11:07
> To: Mass spectrometry standard development
> Cc: psi...@li...; Daniel Schober; psi...@li...
> Subject: Re: [Psidev-ms-vocab] [Psidev-ms-dev] Release candidate 3.74.0_rc1 of psi-ms.obo
>
> Hi,
>
> On Fr, 2015-03-06 at 17:03 +0100, mayerg97 wrote:
> ...
>> Changed CV terms in version 3.74.0_rc1 of psi-ms.obo:
>> =====================================================
>> ************ Changed is_a relationship from 'quality estimation method
>> details'
>> ************ --> 'statistical threshold'
>> [Term]
>> id: MS:1001494
>> name: no threshold
>> def: "In case no threshold was used." [PSI:PI]
>> is_a: MS:1002482 ! statistical threshold
> There are some ontology guidelines, which strongly discourage the uses of negative Terms "no xxx" . In practice, they are sometimes added for pragmatic reasons, but why not leaving blank if no threshold is used ? Since this is a "Changed term", it must have been present before, but I just wanted to mention for the future.
>
> Yours,
> Steffen
>
>
>
>

-- 

Dr. Daniel Schober
Leibniz Institute of Plant Biochemistry, http://www.IPB-Halle.de
Dept. for Stress and Developmental Biology Bioinformatics & Mass Spectrometry
Weinberg 3 Tel. +49 (0) 345 5582 - 1476
06120 Halle