[Psidev-ms-dev] Annotating the CVParam 'value' attribute content in the OBO file of the CV

[Lennart M. <len...@eb...>]

Dear PSI-MS Enthousiasts,


I have had some help from Richard Cote (who developed the Ontology 
Lookup Service and therefore knows about these things) with regards to 
our plan to encode the allowed value type for a CV Param in the PSI-MS 
CV, along with the term.

Apparently (and perhaps not surprisingly), we're not the first to think 
of this :). A nice precedent has been set by the PSI-MOD (Ontology for 
protein modifications). An example is given below:

[Term]
id: MOD:00002
name: O-glycosyl-L-serine
def: "a protein modification that effectively forms an 
O3-glycosylserine" [PSI-MOD:ref]
exact_synonym: "O-glycosylserine" []
exact_synonym: "OGlycoSer" [PSI-MOD:unique short name]
xref_analog: DiffAvg: "146.14"
xref_analog: DiffFormula: "C 6 H 10 N 0 O 4"
xref_analog: DiffMono: "146.057909"
xref_analog: Formula: "C 9 H 15 N 1 O 6"
xref_analog: MassAvg: "233.22"
xref_analog: MassMono: "233.089937"
xref_analog: Origin: "S"
xref_analog: Source: "Natural"
xref_analog: TermSpec:"none"
is_a: MOD:00396 ! O-glycosylated residue
is_a: MOD:00916 ! modified L-serine residue


As you can see, the 'xref_analog' fields are used to encode a variety of 
'annotations' (which is incidentally the way the OLS has been rigged to 
interpret them).

So essentially, the 'xref_analog' stuff for PSI-MS can become something 
like this:

xref_analog: value-type:xsd\:int "The allowed value-type for this CV term."

Note the escaped ':' between xsd and int (OBO uses ':' as an internal 
separator, so it escapes them in values and such).

This mechanism also works beautifully with OBO-Edit (the premier 
graphical OBO file format editing tool), as these 'xref_analog' fields 
are Dbxrefs in the interface, in our example case with database name 
'value-type' and database id 'xsd:int', and comment 'The allowed 
value-type for this CV term.'.

So we can reliably use this proven mechanism. Another potentially nice 
thing is that one term can have many 'xref_analog' fields, so we can 
potentially permit more than one data type (if we would ever want to do 
that).



Cheers,

lnnrt.