Re: [Psidev-ms-dev] PSIDocProcSolicitedReviewerComments

[Steffen N. <sne...@ip...>]

Hi,

here's my 2c:


> Reviewer 1:
...
> Section 1.3.1: The importance of retention time seems to be
> underplayed here, as a transition is defined as a precursor and

Right, I remember discussions here on normalized/predicted/local 
retention times, and their roles should be mentioned in 1.3 

> Section 1.3.2: The meaning and relevance of the last sentence in this
...
> good to have a clear explanation of why Targeted MS/MS is explicitly
> supported in TraML because to me it seems like an unnecessary
> bolt-on. 

I am/was one of the supporters of Targeted MS/MS, 
because we use QqTOF machines where you can't do 
the "classic" MRM experiments. Nevertheless, as the reviewer 
mentioned, the format could be adopted by "Control software 
for mass spectrometry based instruments", and why shouldn't that 
also include targeted MS/MS, and expand beyond "instruments capable 
of carrying out SRM". 

> Sections 1.3.5 and 1.3.6: Neutral loss scans and n>=3 experiments are
> not supported, with good reason. However, it would be good to have
> some reassurance that it would be technically possible to add these
> through a minor version change should the need arise. This would give
> developers additional confidence in adopting the standard.
Actually, this supports my opinion on the inclusion 
of Targeted MS/MS above!

> Section 2.2: Reuse of existing PSI work is great to see. It clearly
> decreased the time needed to design TraML and should similarly
> decrease the time needed to implement it. Bullet point 3, about the
> relationship between TraML and mzIdentML contains some speculation
> about the future. If such speculation is to be included would it not
> be wise to consider that linkage between TraML and a future
> quantitation data standard would be a natural progression?
Maybe that is too much speculation, the document is a *specification*,
rather than speculation ... OTOH, the interplay of the formats 
could be placed in the 1.3 use cases section ?

> Section 2.3.2: I can’t conceive of the special case described where a
> transition would consist solely of two masses and a retention time,
What if there is no identification yet, but somehow a prior experiment 
has shown something to be differential and hence of interest 
for an MRM run ? Actually, why is this section limited 
to "retention times in <Transition>" ? I think it holds true 
for <Target>s as well.

> Reviewer 2:
> M/Z window width
...
>  m/z’ (MS:1000827) and ‘isolation width’ (MS:1000023) are very ion
> trap specific
Nope. The corresponding element in a Bruker micrOTOFq 
method file is "MSMSManual_ListIsolationWidth" for a targeted MSMS run.

> Pure schema versus pure controlled vocabulary
...
> The TraML format relies too heavily on controlled vocabulary. I can
> not think of any practical scenario where the precursor or product of
> a transition would not contain an m/z. Including this one element
> would vastly simplify implementing reading and writing from TraML
> files, and remove dependency on 3rd party tools.
The PSI cvTerm is much more than "just a CV", it's an ontology.
To use it to its full power will require the use 
of a 3rd party tool anyway.

> I’m very pleased to see that a Windows dll from ProteoWizard is
> available to simplify working with TraML files. I would prefer to have
> this dll available directly from the group responsible for TraML, so
> that updates to the TraML format or controlled vocabulary will be
> supported. At a minimum, I would like to see stronger assurances that
> ProteoWizard will be the reference implementation for TraML, and how
> this will be synchronized with TraML updates.
That's beyond the scope of a specification document. 
But the website could do a better job of marketing 
the TraML ecosystem. Do we have any Java Implementations for TraML ?

> Simple format
I guess that format could only handle a limited subset 
of the TraML spec. Section 1.2 mentions that existing CSV stuff 
is very diverse, so why add another flavor ? Maybe pwiz 
could promote one or two flavors to de-facto standards 
by adding converters, but I doubt it makes sense within TraML.

> Reviewer 3:
...
> A few points:
> 1. I was surprised by the decision to exclude neutral loss ions. For
> modified peptides these seem to be key, and need to be considered
> almost always.
Does that mean neutral loss scans ? Or how could those 
be consideren in an MRM experiment ? 

> 2. The decision to keep retention time linked with transition seems is
This was raised by Reviewr #1 as well, see comments above

> allows for possible errors, e.g., allowing for different retention
> time for each peptide.
That could be caught by a validator ?

> Reviewer 4:
> 1) Unless the TraML format is supported by the various vendors and
...
> We didn't see any converters or conversion examples for TraML format
> documented at the site: http://www.psidev.info/index.php?q=node/405.
Again, marketing the ecosystem on the web site.

Yours,
Steffen

-- 
IPB Halle                    AG Massenspektrometrie & Bioinformatik
Dr. Steffen Neumann          http://www.IPB-Halle.DE
Weinberg 3                   http://msbi.bic-gh.de
06120 Halle                  Tel. +49 (0) 345 5582 - 1470
                                  +49 (0) 345 5582 - 0
sneumann(at)IPB-Halle.DE     Fax. +49 (0) 345 5582 - 1409