Re: [Psidev-ms-dev] mzML's binary arrays

[Marc S. <st...@in...>]

Hi Eric,

i agree that less custom encoding is desirable to make files exchangable
between different tools. However in this case i would go both ways,
depending on the significance of the term.

(1)
I use these arrays to store debug information of an algorithm and add
quite a few arrays, depending on the charge states i look at:

pattern_score_1
pattern_score_2
pattern_score_... (one per charge)
intensity_score
local_maximum
trace_score

Putting the name in the userParam is not a good idea because it makes
these arrays unusable for other tools - too custom in my opinion.
Adding a CV term for each such debug variable would however be too much.
So i think the intermediate way is just right:
Terms which are not generally usable should be put to a 'named custom
array'. This would correpond to an optional XML attribute 'name' for the
'binaryDataArray' tag.
But we have to state clearly in the documentation that for more general
terms, a CV entry should be added.

(2)
After peak picking we store much more information than the position and
intensity. The arrays there are:

SignalToNoise
fwhm
leftWidth
rightWidth
maximumIntensity
peakShape
rValue

'SignalToNoise' is alread a CV term. 'fwhm' would be a good candidate
for a CV term as well.
The rest is more algorihtm-dependent and no general concept which is why
we could simply store them in a 'named custom array'.

What do you think?

Best,
Marc

> Hi Marc, I think we would be better off creating CV terms for all the
> kinds of arrays people want to encode. So I'm much rather get a request
> that someone's software wants to write out "full width at half maximum"
> and create a term, furnish an accession number, and thereby publicly let
> all writer and reader authors know that this is a legal entity that
> could occur. No schema change is necessary.
>
> I find this preferable to having a vague slot that could be filled with
>
> full width at half maximum
> full width at half max
> FWHM
>
> in an uncontrolled and variable way.
>
> This is our general aim for mzML. We would like to steer away from
> custom ways of encoding data as much as possible.
>
> Does that seem reasonable?
>
> Would you like "full width at half maximum" to be added to the CV?