[Psidev-ms-dev] requirements survey results

[Kent L. <Ken...@ge...>]

The responses to the PSI-MS requirements survery sent out some time ago =
are aggregated below.
=20
Regards,
Kent Laursen
GenoLogics
=20
PSI-MS Requirements Questionnaire

=20

=20

Requirements

=20

At the most technical level of detail the deliverables of PSI-MS will =
include a codified rendering of information interchange structure in the =
form of UML models and/or XML Schema that can be utilized as a basis for =
a variety of computerized implementations.  However PSI-MS has the goal =
to publish its work at three levels of abstraction in order clearly =
represent its value to a broader audience ranging from scientists and =
decision makers to technologists. These three levels of abstraction are =
the conceptual, the logical and the physical.  This questionnaire asks =
you to categorize your ideas for requirements with this leveling =
applied.  When collated your responses will be a valued input in =
describing mzData in both specification and codified form.  Some =
characteristics of these three levels of abstraction are listed below:

=20

Conceptual =96 visionary, science driven, computationally independent, =
expressing an aim, justifying and or explaining the basis for the effort

=20

Logical - expressed clearly in human understandable terms, logically =
complete without implementation dependence, providing satisfactory =
rendering of what the solution is, delineating the contract(s) that the =
solution must fulfill

=20

Physical =96 having a direct relationship to a codified or machine =
processable form, describing detail sufficient for implementation

=20

We utilize the notion of requirements to simply state a set of aims, =
qualities, rules, constraints, shapes and functionality to which the =
published work of PSI-MS must adhere.   In addition PSI-MS will provide =
a response or mapping describing how its publications address the =
requirements and concerns represented by MCP and MIAPE.=20

=20

=20

=20

Questions

=20

1. What is your aim in participating with the PSI-MS working group

=20

*	Get an in-depth understanding and early information of how the =
proposed standards are to be used.
*	Contribute to the standards by sharing of our expertise in proteomics =
experimentation and data analysis.
*	To contribute with our experience in using the proposed standards in =
software developed by us.
*	We have adopted mzData as our internal mass spec data format and have =
developed tools to convert other ms data formats - eg, mzXML - to it.  I =
participate primarily as an interested consumer of the standards =
developed by the working group.
*	We have interests in ms data that go beyond proteomics.  I would like =
to add a voice to the discussions that might allow the standard to =
encompass a broader discipline, eg glycoproteomics.
*	The creation of data interchange formats and utilities which enhance =
access to raw data for the development and use of informatics tools.  My =
research depends on high throughput and automated mass spectrometry and =
easy access to a full representation of data from instruments is =
critical.
*	To participate in establishing guidelines, defining and evaluating =
object models and derived schemes, and elaborating terminological =
resources in the context of existing initiatives including PSI
*	I am responsible for the standardization of the proteomics platform in =
our research center. Therefore, I am interested in the advancement of =
PSI-MS work concerning mzData and analysisXML.

=20

2. What would you like to see PSI-MS accomplish or stated another way =
what should be the goals of PSI-MS in your view?



*	To create data transport standards =96 for publication repositories, =
between software and database applications.
*	Create guidelines on what data should be retrieved and stored for easy =
access (call it an object model if you will) in proteomics =
experimentation.
*	Make vendors understand that proprietary data formats are not a key =
for success.
*	Define a standard that can capture ms data acquired by most models of =
ms instruments.
Define a standard that can capture the semantics of the ms data within =
any of numerous discipline contexts.
The first standard provides for the raw data, devoid of contextual =
semantics.  The second adds the contextual semantics
*	The goal should be to create a community standard which addresses the =
needs of as many stakeholders as possible.  To me this means recognizing =
several key facts: a) instrument vendors own their data formats and an =
interchange format is not the same as a operational or native format; b) =
there is no single, correct proteomics or mass spectrometry =91work =
flow=92 and as many of these uses need to be considered as possible for =
an interchange specification to become a standard.
*	From the data producer=92s point of view, to ensure common practices =
by sharing experimental data and results thru publications (i.e. =
supplementary materials) and repositories, and finally to be able to =
understand unambiguously and to reproduce proteomics experiments.
*	From the software engineer=92s point of view, to deliver standard =
format (and associated handling tools) that encompasses the following =
types of data: raw, processed and identification. To make it daily used, =
the issued standards might be open and well-documented; moreover, widely =
accepted and supported (especially by vendors and manufacturers).
*	In order to make people use PSI-MS standards, PSI-MS should not only =
define them but also provide users with free open source software tools =
such as MS spectra viewers, MS analysis viewers, parsers, converters, =
=85

=20

3. If you have any general suggestions as to how to accomplish these =
goals, please state them here.

 =20

*	What I think is lacking in PSI-MS is the lack of communication of =
progress and changes in the development. The PSI-dev web pages should be =
a natural place to look for the current status of PSI-MS development. As =
it is now, I get the feeling that a small group knows where PSI-MS is =
going. Since the bi-weekly teleconfs are not scalable to a large =
audience, information must be spread in some other way. I think this is =
a key issue if we want the community to take our work seriously, and in =
the long run adopting standards created by us.
*	Incremental revisions of standards, i.e. do not try to solve =
everything in one shot. Also, be more pragmatic, tests with real data =
during specification development. As it is now, we have problems knowing =
what should be changed in the proposed standards.
*	Get funding to key people so they can focus on PSI-MS
*	Concentrate first on the raw data standard.  You should be able to =
achieve widespread agreement on it rather quickly.  The bun-fight starts =
when the contextual semantics standard is attempted.  This is because =
each research group will have their own context and will want it =
represented.  If the contextual semantics standard can be made abstract =
enough, individual contextual subclasses will be straightforward to =
develop and most groups will be happy.
*	Define the process by which decisions are made.  Formalize the =
structure of the organization and give everyone a chance to contribute.  =
Provide better public interaction, communication and documentation
*	it seems fruitful to allow wide participation and create a transparent =
process for establishing standards. The main point remains the barrier =
of language which can lead to misunderstanding and sometimes, make it =
difficult to explicit semantic concept. For these reasons, frequent =
exchange by mailing list and conf call reports should ensure a good =
communication level between each participant. Besides, document revision =
tracking would be favored by making it frequently available. Two =
meetings per year seem to be enough

=20

4. What conceptual requirements should be fulfilled by PSI-MS/mzData?


=20

*	I think the most important goal for PSI-MS specifications is that they =
should support and enable sharing of data.
*	It should be possible to answer the question =93I=92ve done an =
experiment with 67 samples, MS and MSMS. I need to send these to a =
publication repository, how do I use PSI-MS stuff?=94
*	The formats of the PSI should allow experimental quality to be =
evaluated.
*	They should allow scientists to share information across instrument =
types, time and space.
*	The formats of the PSI should allow the accumulation of a community =
wisdom represented by the ability to reanalyze data indefinitely.

=20

5. What logical requirements should be fulfilled by PSI-MS/mzData?

=20

*	The standards should be implement-able, i.e. don=92t overdo them.
*	Examples of use of the standard should be created, i.e. real data in =
files following specifications created by PSI-MS. Use cases describing =
real world scenarios.
*	The aim of any standard is to promote interchange.  It is also to =
promote computability.  By creating, promoting, and enforcing a =
standard, researchers will be able to share data, build on previous =
findings, and construct useful databases and archives.
*	Something akin to the W3C (HYPERLINK "http://www.w3.org/"www.w3.org).
*	It should not break when new instruments or experiments are developed.
*	mzData should perform the function of being a full representation of =
data from mass spectrometry-based experiments.
*	The formats should be able to be generated by the instrument data =
systems directly and used by users without expertise in programming to =
submit data to repositories, or to interchange results between groups.
*	The solution should be the unique reference: produced by raw data =
processing software as well as identification (search) engines.=20
*	The solution must be pragmatic: for instance, xsl transformation from =
a given format to another one is time- and space-consuming, therefore =
unrealistic in a =93true=94 proteomics pipeline. Another example is =
illustrated by the amount of data daily produced by proteomics center =
(in our case around 30 Go/day of raw data): a human-readable format =
seems to be inappropriate for such volume.

=20

6. What physical requirements should be fulfilled by PSI-MS/mzData?

=20

*	Use LSID to reference data residing in known data providers, e.g. when =
a Swissprot entry is references do something like =
urn:lsid:expasy.ch:swissprot:p12345:41, and this should be possible to =
validate with an parser. This would also remove the need for storing =
information such as sequences, since these will be automatically =
retrievable. Bottom line: embrace LSID.
*	The PSI-MS working group thanks you for your participation in this =
survey.  Your entries will be examined, collated and included in the =
shaping of the published specifications of mzData and analysisXML.
*	Development and maintenance of a schema, eg .xsd, or collection =
thereof
*	Development of tools that can be used to move data from an earlier =
version of the schema to a later version
*	Development of documentation that helps people build datasets that =
conform to the schema
*	Maintenance of a repository of open source tools that have been =
developed by PSI or 3rd parties that operate on dataset conforming to =
the schema
*	It should be compact enough to allow reasonable computer systems to =
move and operate on files of the format.
*	The format should be open and self describing.
*	Files in the format should be verifiable, support concepts like =
digital security and allow for the management of data integrity.
*	To answer properly, we just wonder whether it is expected one (and =
only one) implementation (such as XML) or several (XML, SQL, Object =
oriented language) derived from a unique model?
*	As said in 2., for mzData, PSI-MS should do a work similar to the one =
done by Sashimi for mzXML standards cf:  =
http://sashimi.sourceforge.net/software_glossolalia.html.
*	That is provide free converters from native binary formats (RAW, wiff, =
=85) into mzData and parsers to get spectra objects from an mzData file =
in Perl, Java,=85
*	Same for analysisXML: provide converters from common formats: Sequest, =
Mascot, Tandem, =85 into analysisXML.
*	It should also provide a free portable viewer (cf msInspect =
http://proteomics.fhcrc.org/CPAS/Wiki/home/help/page.view?name=3Dviewer) =
to display MS spectra. InSilicos does the job but is available on =
Windows only.


--=20
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.375 / Virus Database: 267.15.1/250 - Release Date: 2/3/2006
=20