[Psidev-pi-dev] Annotating prediction and matching rules for peptide fragments

[Matthew C. <mat...@va...>]

Hi all,

The prediction and matching rules in database search engines are 
arbitrarily complex algorithms and it is quite intractable to come up 
with a file-scope way to specify the algorithm. Thinking again about 
result-scope specification of the predicted and matched ions, it seems 
possible to come up with an optimized format (minimizing markup bloat). 
Is there value in annotating the predicted but missed ions, or is it 
only desirable to annotate the matched ions? Also, is there an 
established format for specifying the composition of peptide fragments 
(accounting for losses, radicals, and charge)?

There was some suggestion on the last call of referring to a script 
which could run the algorithm. I think that's a reasonable approach if 
the algorithms are put in the CV and there is a version parameter when 
they are referenced by a results file. I would also hope that we 
maintain a link to the script in the CV definition. Possibly, the 
version would have to be part of the term itself, or there could be some 
convention where the script provides a version parameter? Given these 
nasty versioning issues, the result-scope approach starts to look better...

-Matt