Re: [Psidev-pi-dev] FW: Spec doc

[Pierre-Alain B. <pie...@is...>]

Yes, there are tools using more complex rules than only one global metrics:

Phenyx uses this for the fragments in some scoring models (although not 
specified by the user),
Aldente has the option for PMF
Profound proposes for PMF two separate tolerances for data submitted 
with monoisotopic or average precision (a combination is possible)
MSTag (from ProteinProspector) has a "systematic error" value with the 
same unit as the parent tolerance one
MS-Fit has a similar option
Mascot allows "local" tolerance values if the input file is mgf (at 
least the format allows to specify specific tolerances for a given spectrum)

I believe that now we have to live with a first version that considers 
only one simple metrics, particularly if including the complexity would 
require a schema change.

Pierre-Alain


Andreas Bertsch wrote:
> Hi Pierre-Alain,
>
>   
>> Just a question about these tolerances:
>> if one need to specify a union of 2 sets of tolerances, such as 200ppm
>> and 0.1Da (which means 200ppm with a minimum of 0.1Da), can one just
>> have twice the pair of tolerances? for instance:
>>       <FragmentTolerance>
>>         <pf:cvParam accession="PI:00412" name="search tolerance plus value"
>> value="0.5" cvRef="PSI-PI" unitAccession="UO:0000221" unitName="dalton"
>> unitCvRef="UO" />
>>         <pf:cvParam accession="PI:00412" name="search tolerance plus value"
>> value="500" cvRef="PSI-PI" unitAccession="UO:0000221" unitName="ppm"
>> unitCvRef="UO" />
>>         <pf:cvParam accession="PI:00413" name="search tolerance minus
>> value" value="0.5" cvRef="PSI-PI" unitAccession="UO:0000221"
>> unitName="dalton" unitCvRef="UO" />
>>         <pf:cvParam accession="PI:00412" name="search tolerance plus value"
>> value="500" cvRef="PSI-PI" unitAccession="UO:0000221" unitName="ppm"
>> unitCvRef="UO" />
>>       </FragmentTolerance>
>>     
> At the moment this cannot be done. And just repeating the terms would be 
> confusing in my opinion. The meaning that one is a minimum mass tolerance and 
> the other one is the one used if it is greater than the threshold would not be 
> directly obvious while reading the file. If we want to support such constructs 
> we need schema changes here, I think. Are there any search engines which allow 
> for such a specification of the tolerance?
>
> Cheers,
> A.
>
>
>
>