sed-ml-discuss — Migrating to google groups: https://groups.google.com/d/forum/sed-ml-discuss

Re: [SED-ML-discuss] Standard error codes for SED-ML

[Michael H. <mh...@ca...>]

>    I agree with you to  a large extent - it would be interesting to 
> hear the rationale for SBML's inclusion of error codes in their 
> specification document.

Some background and explanation:

In the SBML specification, they are not error codes, in fact, but rather numbered validation and consistency rules.  The SBML Level 1 (all versions) and Level 2 Version 1 specifications don't have them; they were introduced later on, as more people worked from the SBML specifications and it became clear a more systematic approach to defining "validity" was needed.

The problem we tried to solve is rooted in the fact that the schemas do not capture everything about what constitutes a valid SBML document.  A lot is expressed in the text of the specification, but then it's difficult to glean the complete set of requirements by reading the text.  So we tried to list them all explicitly.  

Once you start listing rules like that, it's only natural to number each rule, and to pick numbers that might be compatible with software error codes.

But fundamentally, they would be in the specification even if no software system ever output them.  Although we originally dealt with XML Schema and its limitations, I suspect that even today, with a combination of RELAXNG and Schematron, we wouldn't be able to encode everything in schema rules.  (I could be wrong about that.  We'll know eventually, since we're working on that.)  So, there will probably always be things that can only be encoded as separate algorithmic checks.  Given that's the case, we want to state the rules as clearly as possible.

Best regards,
MH

Re: [SED-ML-discuss] Standard error codes for SED-ML

[Richard A. <ric...@ed...>]

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

Re: [SED-ML-discuss] Standard error codes for SED-ML

[Andrew M. <ak....@au...>]

On 01/11/11 02:03, Richard Adams wrote:
> Hello All,
>
> At present when a SED-ML file is validated we don't have standardised
> categories and error codes for invalid documents, like SBML does ( see
> http://sbml.org/Facilities/Documentation/Error_Categories and Appendix A
> of SBMLl3v1 spec ).
>
> Given that there are likely to be more SED-ML implementations in the
> near future it would be useful to use a defined set of errors in SED-ML.
> It would help client software to standardise support for validation
> errors independent of the SED-ML implementation it was using.
>
> I'd like to add this in as a feature request if there's no overwhelming
> opposition,

Hi Richard,

I think this is undoubtedly a software implementation interface 
specification issue and not an issue directly related to the exchange 
format.

XPSEDAPI is an effort to produce a standard interface between SED-ML 
implementations and tools that use SED-ML (see 
http://sourceforge.net/projects/xpsedapi/) so adding validation 
facilities and error reporting to XPSEDAPI would probably be the most 
appropriate place for it.

Including information about standard interfaces in a format 
specification is not entirely unheard of - for example, the MathML 
specifications include the MathML DOM specification in an appendix, and 
a number of other W3C specifications do the same.

However, I think we want to be consistent: either include all of 
XPSEDAPI in the Appendix, or no information on standardised interfaces; 
there is little point in arbitrarily picking small parts of a standard 
interface to describe, but leaving out the rest.

Before this would make sense, however, I think we would want to see a 
greater level of community uptake for XPSEDAPI (currently, there is only 
one implementation). Therefore, in the short term, I would suggest that 
working on improving XPSEDAPI so it provides a more comprehensive 
interface, and is at least partly supported by several tools, would get 
us to the point where we can consider including XPSEDAPI in an appendix.

Best wishes,
Andrew


>
> Best wishes
> Richard
>
> Dr Richard Adams
> Software Development Team Leader,
> Centre For Systems Biology Edinburgh
> University of Edinburgh
> Tel: 0131 651 9019
> email : ric...@ed... <mailto:ric...@ed...>
> Web: http://csbe.bio.ed.ac.uk/adams.php
>
>
>
>
>
>
>
> The University of Edinburgh is a charitable body, registered in
> Scotland, with registration number SC005336.
>
>
>
> ------------------------------------------------------------------------------
> Get your Android app more play: Bring it to the BlackBerry PlayBook
> in minutes. BlackBerry App World&#153; now supports Android&#153; Apps
> for the BlackBerry&reg; PlayBook&#153;. Discover just how easy and simple
> it is! http://p.sf.net/sfu/android-dev2dev
>
>
>
> _______________________________________________
> SED-ML-discuss mailing list
> SED...@li...
> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss

[SED-ML-discuss] Standard error codes for SED-ML

[Richard A. <ric...@ed...>]

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

[SED-ML-discuss] sed-ml.org downtime

[Nicolas Le N. <le...@eb...>]

Dear colleagues,

Due to scheduled maintenance at the EBI data centre, a complete shutdown 
will be taking place from Friday, 21st October 2011 (14:00 GMT) until the 
evening of Sunday, 23rd October 2011. This will affect access to all 
services hosted at the EBI and not yet moved to the London data centres.

Unfortunately this includes sed-ml.org. We hope to resume services as soon 
as possible after the restart of the systems.

We are very sorry for any inconvenience caused.

Best regards,

-- 
Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/

Re: [SED-ML-discuss] Question about model IDs

[Richard A. <ric...@ed...>]

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

Re: [SED-ML-discuss] Question about model IDs

[David N. <dav...@gm...>]

> The naive solution would be to stick in the type of repository into the URI,
> however since the repositories can use different protocols this might be
> tough. Unless we use a syntax like:
> svn+<actual repository url>
> hg+<actual repository url>
> git+<actual repository url>
> Thoughts?
>
> Well, for the 'model developer' user group, presumably they would have
> access to a project repository. I think adding in authentication info is
> outside the scope of SED-ML. At least on SF, you can just get the
> appropriate revision as a regular URL, e.g.,
> http://sed-ml.svn.sourceforge.net/viewvc/sed-ml/sed-ml/schema/level1/version1/sed-ml-L1-V1.xsd?revision=410&pathrev=410
> I just thought this might reconcile the disparate needs of model developers
> versus model end-users, but perhaps this mechanism is too simplistic.
>
> Correct, the URL you post above would work, however it would not really tell
> the program that you are dealing with an SVN repository. So if we changed
> the protocol from http to svn+http or git+http then tools could know, ok
> this is actually in a repository and i should use those repository routines
> (for example, clone the repository which also would allow for modifications
> … ).

I think this is well outside the scope of SED-ML. If you start having
SED-ML tools needing to know about all the different version control
and remote access protocols then you're back at the problem Nicolas
raised regarding turning off tool developers due to the complexity of
supporting the standard. The overhead of tools only needing to know
about http/https and local files seems to be the right level to aim
at. Someone simply using the SED-ML document is unlike to care how the
models are accessed as long as the simulation experiment runs, and
anyone interested in developing the underlying models is likely to
want to manage the versioning themselves.


Cheers,
David.

Re: [SED-ML-discuss] Question about the sedml zip file

[Herbert S. <hs...@gm...>]

But I (or rather my users) need actual values not just the mean and std.

I am not familiar with NuML, do you have  link?

Herbert

On Fri, Oct 7, 2011 at 9:14 AM, Nicolas Le Novère <le...@eb...> wrote:
> On 07/10/11 17:05, Herbert Sauro wrote:
>> You're right that the different parameter sets could be from a
>> population study and in this sense there is an underlying probability
>> distribution which one would like to record. However can the distrib
>> package represent actual data sets (the values themselves) rather than
>> their statistical distribution?
>
> Yes. This is what I called particle representations. During the initial
> discussions in the June meeting, we used the uncertML RandomSample to do
> that (http://sbml.org/images/c/cc/Eg5.xml), where you can also weight the
> different values.
>
> But this is very much in flux, in particular since the proposal of NuML at
> COMBINE 2011.
>
>> What happens if I have 5 parameters
>> sets with widely different values perhaps obtained from 5 quite
>> different different strains of ecoli.  I don't want to represent the
>> set using statistical summary measures but the actual values
>> themselves, would I use the distrib package in this case?
>
> I believe so, yes. You will also be able to describe a set of values by
> their geometry mean and std.
>
>> Herbert
>>
>>
>> On Fri, Oct 7, 2011 at 12:30 AM, Nicolas Le Novère<le...@eb...>  wrote:
>>> On 06/10/11 23:51, Herbert Sauro wrote:
>>>> Nicolas, I looked up the SBML distrib package but unless I am mistaken
>>>> that seems to be related to specifying statistical data on parameters,
>>>> see
>>>>
>>>> http://sbml.org/Community/Wiki/SBML_Level_3_Proposals/Distributions_and_Ranges
>>>>
>>>> I want to be able to represent multiple parameter sets for a given
>>>> model, preferably named and with descriptions.
>>>
>>> What you describe is a statistical distribution. The distrib package will
>>> support particle representations. One of the main targets of the package
>>> will be pharmacometrics models, where the model/data is often represented
>>> by an ensemble of patient data.
>>>
>>>> How should I deal with this situation: Multiple parameter sets for a
>>>> given model where each parameter set needs a slightly different
>>>> simulation, eg different time end for a continuous simulation.
>>>
>>> Ouch. In the case where both model and simulation descriptions are
>>> affected, I agree that this belong to SED-ML. You'll need a forthcoming
>>> version with NuML integrated.
>>>
>>>>
>>>> Herbert
>>>>
>>>>>  From 33K feet somewhere above continental USA.
>>>>
>>>> ------------------------------------------------------------------------------
>>>> All the data continuously generated in your IT infrastructure contains a
>>>> definitive record of customers, application performance, security
>>>> threats, fraudulent activity and more. Splunk takes this data and makes
>>>> sense of it. Business sense. IT sense. Common sense.
>>>> http://p.sf.net/sfu/splunk-d2dcopy1
>>>> _______________________________________________
>>>> SED-ML-discuss mailing list
>>>> SED...@li...
>>>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>>>
>>>
>>> --
>>> Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
>>> Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
>>> Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
>>> http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> All of the data generated in your IT infrastructure is seriously valuable.
>>> Why? It contains a definitive record of application performance, security
>>> threats, fraudulent activity, and more. Splunk takes this data and makes
>>> sense of it. IT sense. And common sense.
>>> http://p.sf.net/sfu/splunk-d2dcopy2
>>> _______________________________________________
>>> SED-ML-discuss mailing list
>>> SED...@li...
>>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>>>
>>
>> ------------------------------------------------------------------------------
>> All of the data generated in your IT infrastructure is seriously valuable.
>> Why? It contains a definitive record of application performance, security
>> threats, fraudulent activity, and more. Splunk takes this data and makes
>> sense of it. IT sense. And common sense.
>> http://p.sf.net/sfu/splunk-d2dcopy2
>> _______________________________________________
>> SED-ML-discuss mailing list
>> SED...@li...
>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>
>
> --
> Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
> Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
> Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
> http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/
>
>
> ------------------------------------------------------------------------------
> All of the data generated in your IT infrastructure is seriously valuable.
> Why? It contains a definitive record of application performance, security
> threats, fraudulent activity, and more. Splunk takes this data and makes
> sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-d2dcopy2
> _______________________________________________
> SED-ML-discuss mailing list
> SED...@li...
> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>

Re: [SED-ML-discuss] Question about the sedml zip file

[Nicolas Le N. <le...@eb...>]

On 07/10/11 17:05, Herbert Sauro wrote:
> You're right that the different parameter sets could be from a
> population study and in this sense there is an underlying probability
> distribution which one would like to record. However can the distrib
> package represent actual data sets (the values themselves) rather than
> their statistical distribution?

Yes. This is what I called particle representations. During the initial 
discussions in the June meeting, we used the uncertML RandomSample to do 
that (http://sbml.org/images/c/cc/Eg5.xml), where you can also weight the 
different values.

But this is very much in flux, in particular since the proposal of NuML at 
COMBINE 2011.

> What happens if I have 5 parameters
> sets with widely different values perhaps obtained from 5 quite
> different different strains of ecoli.  I don't want to represent the
> set using statistical summary measures but the actual values
> themselves, would I use the distrib package in this case?

I believe so, yes. You will also be able to describe a set of values by 
their geometry mean and std.

> Herbert
>
>
> On Fri, Oct 7, 2011 at 12:30 AM, Nicolas Le Novère<le...@eb...>  wrote:
>> On 06/10/11 23:51, Herbert Sauro wrote:
>>> Nicolas, I looked up the SBML distrib package but unless I am mistaken
>>> that seems to be related to specifying statistical data on parameters,
>>> see
>>>
>>> http://sbml.org/Community/Wiki/SBML_Level_3_Proposals/Distributions_and_Ranges
>>>
>>> I want to be able to represent multiple parameter sets for a given
>>> model, preferably named and with descriptions.
>>
>> What you describe is a statistical distribution. The distrib package will
>> support particle representations. One of the main targets of the package
>> will be pharmacometrics models, where the model/data is often represented
>> by an ensemble of patient data.
>>
>>> How should I deal with this situation: Multiple parameter sets for a
>>> given model where each parameter set needs a slightly different
>>> simulation, eg different time end for a continuous simulation.
>>
>> Ouch. In the case where both model and simulation descriptions are
>> affected, I agree that this belong to SED-ML. You'll need a forthcoming
>> version with NuML integrated.
>>
>>>
>>> Herbert
>>>
>>>>  From 33K feet somewhere above continental USA.
>>>
>>> ------------------------------------------------------------------------------
>>> All the data continuously generated in your IT infrastructure contains a
>>> definitive record of customers, application performance, security
>>> threats, fraudulent activity and more. Splunk takes this data and makes
>>> sense of it. Business sense. IT sense. Common sense.
>>> http://p.sf.net/sfu/splunk-d2dcopy1
>>> _______________________________________________
>>> SED-ML-discuss mailing list
>>> SED...@li...
>>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>>
>>
>> --
>> Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
>> Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
>> Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
>> http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/
>>
>>
>> ------------------------------------------------------------------------------
>> All of the data generated in your IT infrastructure is seriously valuable.
>> Why? It contains a definitive record of application performance, security
>> threats, fraudulent activity, and more. Splunk takes this data and makes
>> sense of it. IT sense. And common sense.
>> http://p.sf.net/sfu/splunk-d2dcopy2
>> _______________________________________________
>> SED-ML-discuss mailing list
>> SED...@li...
>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>>
>
> ------------------------------------------------------------------------------
> All of the data generated in your IT infrastructure is seriously valuable.
> Why? It contains a definitive record of application performance, security
> threats, fraudulent activity, and more. Splunk takes this data and makes
> sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-d2dcopy2
> _______________________________________________
> SED-ML-discuss mailing list
> SED...@li...
> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss


-- 
Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/

Re: [SED-ML-discuss] Question about the sedml zip file

[Herbert S. <hs...@gm...>]

You're right that the different parameter sets could be from a
population study and in this sense there is an underlying probability
distribution which one would like to record. However can the distrib
package represent actual data sets (the values themselves) rather than
their statistical distribution? What happens if I have 5 parameters
sets with widely different values perhaps obtained from 5 quite
different different strains of ecoli.  I don't want to represent the
set using statistical summary measures but the actual values
themselves, would I use the distrib package in this case?

Herbert


On Fri, Oct 7, 2011 at 12:30 AM, Nicolas Le Novère <le...@eb...> wrote:
> On 06/10/11 23:51, Herbert Sauro wrote:
>> Nicolas, I looked up the SBML distrib package but unless I am mistaken
>> that seems to be related to specifying statistical data on parameters,
>> see
>>
>> http://sbml.org/Community/Wiki/SBML_Level_3_Proposals/Distributions_and_Ranges
>>
>> I want to be able to represent multiple parameter sets for a given
>> model, preferably named and with descriptions.
>
> What you describe is a statistical distribution. The distrib package will
> support particle representations. One of the main targets of the package
> will be pharmacometrics models, where the model/data is often represented
> by an ensemble of patient data.
>
>> How should I deal with this situation: Multiple parameter sets for a
>> given model where each parameter set needs a slightly different
>> simulation, eg different time end for a continuous simulation.
>
> Ouch. In the case where both model and simulation descriptions are
> affected, I agree that this belong to SED-ML. You'll need a forthcoming
> version with NuML integrated.
>
>>
>> Herbert
>>
>>> From 33K feet somewhere above continental USA.
>>
>> ------------------------------------------------------------------------------
>> All the data continuously generated in your IT infrastructure contains a
>> definitive record of customers, application performance, security
>> threats, fraudulent activity and more. Splunk takes this data and makes
>> sense of it. Business sense. IT sense. Common sense.
>> http://p.sf.net/sfu/splunk-d2dcopy1
>> _______________________________________________
>> SED-ML-discuss mailing list
>> SED...@li...
>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>
>
> --
> Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
> Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
> Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
> http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/
>
>
> ------------------------------------------------------------------------------
> All of the data generated in your IT infrastructure is seriously valuable.
> Why? It contains a definitive record of application performance, security
> threats, fraudulent activity, and more. Splunk takes this data and makes
> sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-d2dcopy2
> _______________________________________________
> SED-ML-discuss mailing list
> SED...@li...
> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>

Re: [SED-ML-discuss] Question about model IDs

[Frank T. B. <fbe...@ca...>]

>> 
>> The naive solution would be to stick in the type of repository into the URI, however since the repositories can use different protocols this might be tough. Unless we use a syntax like: 
>> 
>> svn+<actual repository url>
>> hg+<actual repository url>
>> git+<actual repository url>
>> 
>> Thoughts?
> 
> 
> Well, for the 'model developer' user group, presumably they would have access to a project repository. I think adding in authentication info is outside the scope of SED-ML. At least on SF, you can just get the appropriate revision as a regular URL, e.g., 
> http://sed-ml.svn.sourceforge.net/viewvc/sed-ml/sed-ml/schema/level1/version1/sed-ml-L1-V1.xsd?revision=410&pathrev=410
> 
> I just thought this might reconcile the disparate needs of model developers versus model end-users, but perhaps this mechanism is too simplistic.
> 

Correct, the URL you post above would work, however it would not really tell the program that you are dealing with an SVN repository. So if we changed the protocol from http to svn+http or git+http then tools could know, ok this is actually in a repository and i should use those repository routines (for example, clone the repository which also would allow for modifications … ). 

Frank

Re: [SED-ML-discuss] Question about model IDs

[Nicolas Le N. <le...@eb...>]

On 07/10/11 16:36, Richard Adams wrote:

>> No matter what, I'd prefer for the archive not to go away.
>
> No, I wasn't suggesting we bin the archive idea altogether, it's actually
> very convenient, and I've developed a fair bit of support for the proposed
> archive in software. I just meant that the archive may not be needed for
> exchange, if all necessary resources are available through a URL.

IMHO it does not matter. An archive could contain only a SED-ML file 
referring to everything through URIs. Or it could contain a SED-ML file 
that refers to a local data-set contained in the archive, and point to a 
model using a URI.

-- 
Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/

Re: [SED-ML-discuss] Question about model IDs

[Richard A. <ric...@ed...>]

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

Re: [SED-ML-discuss] Question about model IDs

[Frank T. B. <fbe...@ca...>]

> Herbert was asking  about SED-ML use cases, we don't really have this information on the SED-ML website but probably should do so, it would help clarify its purpose to newcomers. 
> 

Agreed, do you want to volunteer taking a first stab at the text (perhaps in a google doc) and then we massage it and put it online. 

> David, is there any reason why SED-ML files can't use SVN or mercurial URLs to access the  resources? In this case an archive file isn't really needed - just exchange a SED-ML file, which could itself be under version control within a project.
> 

No matter what, I'd prefer for the archive not to go away. It will always be useful to have a snapshot (including all necessary data / models … ) to pass it along. but I'm actually not against having URLs to SVN / git / mercurial. It is an overhead but if the community decides that it is needed why not. Here the potential issues that would need to be resolved: 

1. we need to identify the type of repository (SVN |  HG | GIT …
2. we might have to deal with authentication (though in a way this could hit us already today with the use of URLs)

The naive solution would be to stick in the type of repository into the URI, however since the repositories can use different protocols this might be tough. Unless we use a syntax like: 

svn+<actual repository url>
hg+<actual repository url>
git+<actual repository url>

Thoughts?
Frank

> Best wishes,
> 
> Richard
> 
> 
> 
> 
> On 6 Oct 2011, at 22:28, David Nickerson wrote:
> 
>> The idea of an archive file works well when you have a single
>> transferable set of data that gets shipped from user to user with no
>> need to track or merge changes from different users (i.e., one
>> developer many users). DOCX archives work because Word provides very
>> sophisticated edit tracking and merging functionality - is this
>> functionality something that SED-ML tools want to develop and support?
>> 
>> An alternative, that we have been using successfully in the Physiome
>> Model Repository, is the idea of using mercurial repositories as a
>> "workspace" in which you contain all data related to a piece of work.
>> This piece of work could be anything from a complete model with
>> simulation descriptions and outputs, experimental data used to fit the
>> model, etc. to a single sub-component of a generic model constituent
>> designed as part of a component library. While we use mercurial, any
>> versioning system could be used instead. The advantage of this
>> approach is that the user is able to embed repositories within
>> repositories (subrepos in mercurial, submodules in git, externals in
>> svn) and manage the versioning of embedded repositories such that the
>> user can choose whether to track latest changes or fix on a specific
>> version that is know to be "correct".
>> 
>> We find that this allows modellers to easily[1] ship their work
>> between each other while maintaining a decent provenance record of the
>> developments. While it is not always going to be optimal to include
>> all types of data in a single type of repository, mercurial at least
>> (and probably the other systems), allows embedding different types of
>> repository within a repository. We're starting now to think about how
>> this would work when you start wanting to include datasets that are
>> huge (e.g., large image collections used to fit patient specific
>> cardiac models, or the simulation results of several million cell
>> models in a stomach). But at least in terms of the work currently
>> being done with CellML models, mercurial  repositories are proving
>> sufficient.
>> 
>> When a user simply wants to grab a particular piece of work from the
>> repository for their own use, the website (and soon webservices)
>> provide a way to download a static archive containing the contents of
>> the workspace. Again, we're still trying to work out what to do in the
>> case where an workspace links to datasets that can easily be multiple
>> gigabytes in size. And we are certainly open to any help on this
>> issue.
>> 
>> Cheers,
>> David.
>> 
>> [1] There is currently a lack of tool support for this, so users
>> pretty much have to manage their mercurial workspaces using mercurial
>> directly. This is not always ideal and is also something we are
>> thinking about and working on to some degree for certain applications.
>> 
>> On Fri, Oct 7, 2011 at 9:41 AM, Richard Adams <ric...@ed...> wrote:
>>> 
>>> For those unfamiliar with the concept of the SEDX archive format: It's a
>>> zipped archive of a SED-ML file and >=1 model file. Would an equally simple
>>> solution be acceptable to incorporate other data files as well?
>>> For example, a zip with subfolders
>>> models/
>>> simulations/
>>> data/
>>> diagrams/
>>> results/
>>> Each resource could refer to other resources via relative URIs. E.g., in the
>>> scenario above, a SedML file would refer to ../models/model.xml. Of course,
>>> within an archive, external public resources can still be referred to if
>>> need be - for example, if a dataset is very large.
>>> In this approach all the folders are 'source' folders (i.e., inputs to
>>> computational tasks) except for results/, which could be used to contain the
>>> results of an experiment if need be.
>>> In SBSI we have the concept of a 'project' for modelling resources, with a
>>> defined folder structure largely similar to that proposed above, to enable
>>> tooling to reliably link and detect  resource files, and allow export of
>>> zipped archives for exchange between collaborators.  I'm not especially
>>> advocating its exact usage but just to say it's worked well for us.
>>> Best wishes
>>> Richard
>>> On 6 Oct 2011, at 17:05, Nicolas Le Novère wrote:
>>> 
>>> It seems to me that the need of an archive is pervasive in all the M&S
>>> efforts. We want the models and the simulation descriptions together, but
>>> also the simulation results or the data-sets used to complement or fit the
>>> models, and of course the SBGN-ML files.
>>> 
>>> And we are going in the same direction in DDMoRe, where the need is clearer
>>> since the mathematics only represent part of the "model".
>>> 
>>> After all the  it is just like ODF or DOCX.
>>> 
>>> I therefore propose to move that discussion to combine-discuss.
>>> 
>>> On 06/10/11 16:43, Herbert Sauro wrote:
>>> 
>>> Thank you Richard, that is exactly what I wanted to know, will read appendix
>>> D.  The zip file idea has been floated around for a while (since 2004 I
>>> think by Cliff Shaffer in particular) but in the past there was never
>>> anything to zip with the sbml model until the advent of sedml.
>>> 
>>> Herbert
>>> 
>>> Sent from my iPad, hence excuse my typos.
>>> 
>>> ------------------------------------------------------------------------------
>>> 
>>> All the data continuously generated in your IT infrastructure contains a
>>> 
>>> definitive record of customers, application performance, security
>>> 
>>> threats, fraudulent activity and more. Splunk takes this data and makes
>>> 
>>> sense of it. Business sense. IT sense. Common sense.
>>> 
>>> http://p.sf.net/sfu/splunk-d2dcopy1
>>> 
>>> _______________________________________________
>>> 
>>> SED-ML-discuss mailing list
>>> 
>>> SED...@li...
>>> 
>>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>>> 
>>> 
>>> --
>>> Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
>>> Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
>>> Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
>>> http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/
>>> 
>>> 
>>> ------------------------------------------------------------------------------
>>> All the data continuously generated in your IT infrastructure contains a
>>> definitive record of customers, application performance, security
>>> threats, fraudulent activity and more. Splunk takes this data and makes
>>> sense of it. Business sense. IT sense. Common sense.
>>> http://p.sf.net/sfu/splunk-d2dcopy1
>>> _______________________________________________
>>> SED-ML-discuss mailing list
>>> SED...@li...
>>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>>> 
>>> 
>>> Dr Richard Adams
>>> Software Development Team Leader,
>>> Centre For Systems Biology Edinburgh
>>> University of Edinburgh
>>> Tel: 0131 651 9019
>>> email : ric...@ed...
>>> Web: http://csbe.bio.ed.ac.uk/adams.php
>>> 
>>> 
>>> 
>>> 
>>> 
>>> The University of Edinburgh is a charitable body, registered in
>>> Scotland, with registration number SC005336.
>>> 
>>> ------------------------------------------------------------------------------
>>> All the data continuously generated in your IT infrastructure contains a
>>> definitive record of customers, application performance, security
>>> threats, fraudulent activity and more. Splunk takes this data and makes
>>> sense of it. Business sense. IT sense. Common sense.
>>> http://p.sf.net/sfu/splunk-d2dcopy1
>>> _______________________________________________
>>> SED-ML-discuss mailing list
>>> SED...@li...
>>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>>> 
>>> 
>> 
>> ------------------------------------------------------------------------------
>> All the data continuously generated in your IT infrastructure contains a
>> definitive record of customers, application performance, security
>> threats, fraudulent activity and more. Splunk takes this data and makes
>> sense of it. Business sense. IT sense. Common sense.
>> http://p.sf.net/sfu/splunk-d2dcopy1
>> _______________________________________________
>> SED-ML-discuss mailing list
>> SED...@li...
>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>> 
> 
> Dr Richard Adams
> Software Development Team Leader,
> Centre For Systems Biology Edinburgh
> University of Edinburgh 
> Tel: 0131 651 9019
> email : ric...@ed...
> Web: http://csbe.bio.ed.ac.uk/adams.php
> 
> 
> 
> 
> 
> The University of Edinburgh is a charitable body, registered in
> Scotland, with registration number SC005336.
> ------------------------------------------------------------------------------
> All of the data generated in your IT infrastructure is seriously valuable.
> Why? It contains a definitive record of application performance, security
> threats, fraudulent activity, and more. Splunk takes this data and makes
> sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-d2dcopy2_______________________________________________
> SED-ML-discuss mailing list
> SED...@li...
> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss

Re: [SED-ML-discuss] Question about model IDs

[Richard A. <ric...@ed...>]

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

Re: [SED-ML-discuss] Question about the sedml zip file

[Richard A. <ric...@ed...>]

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

Re: [SED-ML-discuss] Question about the sedml zip file

[Richard A. <ric...@ed...>]

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

Re: [SED-ML-discuss] Question about the sedml zip file

[Nicolas Le N. <le...@eb...>]

On 07/10/11 00:12, Lucian Smith wrote:
> One of the goals of the distrib package is to allow you to store vectors
> that you choose from instead of single values.  But I *think* that the
> idea is to choose randomly from that vector, not to say 'use the first
> value in all vectors for this simulation, then use the second value in all
> vectors for the second simulation'.  Though I could be mistaken.

I believe you are neither right or wrong. The distrib package, and SBML, 
says nothing about the use of those values. With the distrib package, the 
value of an initial condition can be replaced by a distribution. This can 
be a random variable (including truncated ones, themselves including 
ranges) or a set of values. What you do with that is not SBML's business. I 
think according to the type of analysis or simulation you run, you will 
either use a given value per analysis, or all the values at once. In the 
former case, you may use 1, a sample or all of them in alternative analyses.


-- 
Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/

Re: [SED-ML-discuss] Question about the sedml zip file

[Nicolas Le N. <le...@eb...>]

On 06/10/11 23:51, Herbert Sauro wrote:
> Nicolas, I looked up the SBML distrib package but unless I am mistaken
> that seems to be related to specifying statistical data on parameters,
> see
>
> http://sbml.org/Community/Wiki/SBML_Level_3_Proposals/Distributions_and_Ranges
>
> I want to be able to represent multiple parameter sets for a given
> model, preferably named and with descriptions.

What you describe is a statistical distribution. The distrib package will 
support particle representations. One of the main targets of the package 
will be pharmacometrics models, where the model/data is often represented 
by an ensemble of patient data.

> How should I deal with this situation: Multiple parameter sets for a
> given model where each parameter set needs a slightly different
> simulation, eg different time end for a continuous simulation.

Ouch. In the case where both model and simulation descriptions are 
affected, I agree that this belong to SED-ML. You'll need a forthcoming 
version with NuML integrated.

>
> Herbert
>
>> From 33K feet somewhere above continental USA.
>
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure contains a
> definitive record of customers, application performance, security
> threats, fraudulent activity and more. Splunk takes this data and makes
> sense of it. Business sense. IT sense. Common sense.
> http://p.sf.net/sfu/splunk-d2dcopy1
> _______________________________________________
> SED-ML-discuss mailing list
> SED...@li...
> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss


-- 
Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/

Re: [SED-ML-discuss] Question about the sedml zip file

[Herbert S. <hs...@gm...>]

JDesigner currently supports it own annotation for storing multiple
data sets, has done for many years. I just thought there might be
something on the standards front  that I could use. I would have
thought however that SEDML would have been a convenient place to
reference parameter sets particularly for the use case I gave.
Simulation experiments need data so there has to be some link from
sedml to data sources, be in parameter sets or experimental data for
model fitting, or even statistical information on the the parameters
being used although parameter variability is now in SBML. I am sure
the sedml community has thought of this but I didn't find anything in
the draft to indicate this.

You could also include multiple parameter sets explicitly in Antimony.

Herbert

On Thu, Oct 6, 2011 at 4:12 PM, Lucian Smith <lp...@sp...> wrote:
> One of the goals of the distrib package is to allow you to store vectors
> that you choose from instead of single values.  But I *think* that the
> idea is to choose randomly from that vector, not to say 'use the first
> value in all vectors for this simulation, then use the second value in all
> vectors for the second simulation'.  Though I could be mistaken.
>
> If you want to wait for the 'arrays and sets' package, that might work.
> But it does suffer from the small drawback of not existing.
>
> I can imagine using comp for this, too (naturally).  In Antimony:
>
> model basemod(k1, k2)
>  S1 -> S2 + S3; k1*S1;
>  S2 -> S4;      k2*S2;
> end
>
> model set1()
>  basemod(k1, k2);
>  k1 = 3;
>  k2 = 10;
> end
>
> model set2()
>  basemod(k1, k2);
>  k1 = 10;
>  k2 = 3;
> end
>
> model set3()
>  basemod(k1, k2);
>  k1 = 7;
>  k2 = 7;
> end
>
> etc.
>
> I don't know if you can pick a particular model out of a file for
> simulation in SED-ML?  If so, that'd be all you need; if not, you'd need a
> file per set#() model.  (I wouldn't imagine it would be terribly difficult
> to extend SED-ML to do this, if it doesn't already.)
>
> -Lucian
>
> * Herbert Sauro <hs...@gm...> [2011-10-06 23:52] writes:
>> Nicolas, I looked up the SBML distrib package but unless I am mistaken
>> that seems to be related to specifying statistical data on parameters,
>> see
>>
>> http://sbml.org/Community/Wiki/SBML_Level_3_Proposals/Distributions_and_Ranges
>>
>> I want to be able to represent multiple parameter sets for a given
>> model, preferably named and with descriptions.
>>
>> How should I deal with this situation: Multiple parameter sets for a
>> given model where each parameter set needs a slightly different
>> simulation, eg different time end for a continuous simulation.
>>
>> Herbert
>>
>> >From 33K feet somewhere above continental USA.
>>
>> ------------------------------------------------------------------------------
>> All the data continuously generated in your IT infrastructure contains a
>> definitive record of customers, application performance, security
>> threats, fraudulent activity and more. Splunk takes this data and makes
>> sense of it. Business sense. IT sense. Common sense.
>> http://p.sf.net/sfu/splunk-d2dcopy1
>> _______________________________________________
>> SED-ML-discuss mailing list
>> SED...@li...
>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure contains a
> definitive record of customers, application performance, security
> threats, fraudulent activity and more. Splunk takes this data and makes
> sense of it. Business sense. IT sense. Common sense.
> http://p.sf.net/sfu/splunk-d2dcopy1
> _______________________________________________
> SED-ML-discuss mailing list
> SED...@li...
> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>

Re: [SED-ML-discuss] Question about the sedml zip file

[Lucian S. <lp...@sp...>]

One of the goals of the distrib package is to allow you to store vectors 
that you choose from instead of single values.  But I *think* that the 
idea is to choose randomly from that vector, not to say 'use the first 
value in all vectors for this simulation, then use the second value in all 
vectors for the second simulation'.  Though I could be mistaken.

If you want to wait for the 'arrays and sets' package, that might work.  
But it does suffer from the small drawback of not existing.

I can imagine using comp for this, too (naturally).  In Antimony:

model basemod(k1, k2)
  S1 -> S2 + S3; k1*S1;
  S2 -> S4;      k2*S2;
end

model set1()
  basemod(k1, k2);
  k1 = 3;
  k2 = 10;
end

model set2()
  basemod(k1, k2);
  k1 = 10;
  k2 = 3;
end

model set3()
  basemod(k1, k2);
  k1 = 7;
  k2 = 7;
end

etc.

I don't know if you can pick a particular model out of a file for 
simulation in SED-ML?  If so, that'd be all you need; if not, you'd need a 
file per set#() model.  (I wouldn't imagine it would be terribly difficult 
to extend SED-ML to do this, if it doesn't already.)

-Lucian

* Herbert Sauro <hs...@gm...> [2011-10-06 23:52] writes:
> Nicolas, I looked up the SBML distrib package but unless I am mistaken
> that seems to be related to specifying statistical data on parameters,
> see
> 
> http://sbml.org/Community/Wiki/SBML_Level_3_Proposals/Distributions_and_Ranges
> 
> I want to be able to represent multiple parameter sets for a given
> model, preferably named and with descriptions.
> 
> How should I deal with this situation: Multiple parameter sets for a
> given model where each parameter set needs a slightly different
> simulation, eg different time end for a continuous simulation.
> 
> Herbert
> 
> >From 33K feet somewhere above continental USA.
> 
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure contains a
> definitive record of customers, application performance, security
> threats, fraudulent activity and more. Splunk takes this data and makes
> sense of it. Business sense. IT sense. Common sense.
> http://p.sf.net/sfu/splunk-d2dcopy1
> _______________________________________________
> SED-ML-discuss mailing list
> SED...@li...
> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss

[SED-ML-discuss] Question about the sedml zip file

[Herbert S. <hs...@gm...>]

Nicolas, I looked up the SBML distrib package but unless I am mistaken
that seems to be related to specifying statistical data on parameters,
see

http://sbml.org/Community/Wiki/SBML_Level_3_Proposals/Distributions_and_Ranges

I want to be able to represent multiple parameter sets for a given
model, preferably named and with descriptions.

How should I deal with this situation: Multiple parameter sets for a
given model where each parameter set needs a slightly different
simulation, eg different time end for a continuous simulation.

Herbert

>From 33K feet somewhere above continental USA.

Re: [SED-ML-discuss] Question about model IDs

[David N. <dav...@gm...>]

> I think your solution is neat, and works well for given collaborative
> projects. It is quite hard to put in practice when it comes to distributing
> models around. You cannot force all users to install mercurial (or any
> versioning system), or all tool developers to embed mercurial in their tool.
> One of the big complain BioModels Database receive at the moment from
> non-hardcore people is that they cannot download a file and run it. This is
> what they get with MatLab, and until we deliver the same simplicity of
> usage, they see standards as useless hindrance. The model file can come from
> an online database, be given on a usb key etc.
>
> What we have to develop over the next 2 years must allow the following:
>
> You launch your favourite tool, you import the file, click run and get the
> curves.

you can get this now if you're happy to reference models/data in
online databases as you can easily refer to specific revisions of
models in a repository using URLs. I think the static archive works
well for this particular case when you don't want to assume internet
connectivity, and for the case when the data you need to run a
simulation is not ridiculously huge.

I guess I am coming more from the model development direction, where
it is much more important to track provenance and versioning of all
kinds of data. Tool support is always going to be an issue, but I
think the return on investment for model authors is quite significant,
in this case. I know that I for one am interested to see how things go
once the comp package starts getting widespread usage...


Cheers,
David.

Re: [SED-ML-discuss] Question about model IDs

[Lucian S. <lp...@sp...>]

Hmm, I suppose loading multiple SED-ML files is effectively the same (from 
the point of view of a program that wants to run the simulations and 
display the results) as concatenating them all in one file, isn't it?  
Except for the ID issue.  In both cases you have a bunch of simulations, 
models, and results, and you can either just run everything or you can ask 
a user which bit they want to run.  In that case, I might vote for the 
'standard' of 'what do we include in a SED-ML archive file' to be extended 
to say 'multiple SED-ML files are fine'.

(If indeed it's actually been formally standardized!)

-Lucian

* Richard Adams <ric...@ed...> [2011-10-06 21:52] writes:
> 
> For those unfamiliar with the concept of the SEDX archive format: It's  
> a zipped archive of a SED-ML file and >=1 model file. Would an equally  
> simple solution be acceptable to incorporate other data files as well?
> 
> For example, a zip with subfolders
> 
> models/
> simulations/
> data/
> diagrams/
> results/
> 
> Each resource could refer to other resources via relative URIs. E.g.,  
> in the scenario above, a SedML file would refer to ../models/ 
> model.xml. Of course, within an archive, external public resources can  
> still be referred to if need be - for example, if a dataset is very  
> large.
> In this approach all the folders are 'source' folders (i.e., inputs to  
> computational tasks) except for results/, which could be used to  
> contain the results of an experiment if need be.
> 
> In SBSI we have the concept of a 'project' for modelling resources,  
> with a defined folder structure largely similar to that proposed  
> above, to enable tooling to reliably link and detect  resource files,  
> and allow export of zipped archives for exchange between  
> collaborators.  I'm not especially advocating its exact usage but just  
> to say it's worked well for us.
> 
> Best wishes
> 
> Richard
> 
> On 6 Oct 2011, at 17:05, Nicolas Le Nov?re wrote:
> 
> >It seems to me that the need of an archive is pervasive in all the M&S
> >efforts. We want the models and the simulation descriptions  
> >together, but
> >also the simulation results or the data-sets used to complement or  
> >fit the
> >models, and of course the SBGN-ML files.
> >
> >And we are going in the same direction in DDMoRe, where the need is  
> >clearer
> >since the mathematics only represent part of the "model".
> >
> >After all the  it is just like ODF or DOCX.
> >
> >I therefore propose to move that discussion to combine-discuss.
> >
> >On 06/10/11 16:43, Herbert Sauro wrote:
> >>
> >>Thank you Richard, that is exactly what I wanted to know, will read  
> >>appendix D.  The zip file idea has been floated around for a while  
> >>(since 2004 I think by Cliff Shaffer in particular) but in the past  
> >>there was never anything to zip with the sbml model until the  
> >>advent of sedml.
> >>
> >>Herbert
> >>
> >>Sent from my iPad, hence excuse my typos.
> >>------------------------------------------------------------------------------
> >>All the data continuously generated in your IT infrastructure  
> >>contains a
> >>definitive record of customers, application performance, security
> >>threats, fraudulent activity and more. Splunk takes this data and  
> >>makes
> >>sense of it. Business sense. IT sense. Common sense.
> >>http://p.sf.net/sfu/splunk-d2dcopy1
> >>_______________________________________________
> >>SED-ML-discuss mailing list
> >>SED...@li...
> >>https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
> >
> >
> >-- 
> >Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
> >Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
> >Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
> >http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/
> >
> >
> >------------------------------------------------------------------------------
> >All the data continuously generated in your IT infrastructure  
> >contains a
> >definitive record of customers, application performance, security
> >threats, fraudulent activity and more. Splunk takes this data and  
> >makes
> >sense of it. Business sense. IT sense. Common sense.
> >http://p.sf.net/sfu/splunk-d2dcopy1
> >_______________________________________________
> >SED-ML-discuss mailing list
> >SED...@li...
> >https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
> >
> 
> Dr Richard Adams
> Software Development Team Leader,
> Centre For Systems Biology Edinburgh
> University of Edinburgh
> Tel: 0131 651 9019
> email : ric...@ed...
> Web: http://csbe.bio.ed.ac.uk/adams.php
> 
> 
> 
> 
> 

Content-Description: Edinburgh University charitable status
> The University of Edinburgh is a charitable body, registered in
> Scotland, with registration number SC005336.

> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure contains a
> definitive record of customers, application performance, security
> threats, fraudulent activity and more. Splunk takes this data and makes
> sense of it. Business sense. IT sense. Common sense.
> http://p.sf.net/sfu/splunk-d2dcopy1
> _______________________________________________
> SED-ML-discuss mailing list
> SED...@li...
> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss

Re: [SED-ML-discuss] Question about model IDs

[Nicolas Le N. <le...@eb...>]

David,

I think your solution is neat, and works well for given collaborative 
projects. It is quite hard to put in practice when it comes to distributing 
models around. You cannot force all users to install mercurial (or any 
versioning system), or all tool developers to embed mercurial in their 
tool. One of the big complain BioModels Database receive at the moment from 
non-hardcore people is that they cannot download a file and run it. This is 
what they get with MatLab, and until we deliver the same simplicity of 
usage, they see standards as useless hindrance. The model file can come 
from an online database, be given on a usb key etc.

What we have to develop over the next 2 years must allow the following:

You launch your favourite tool, you import the file, click run and get the 
curves.

On 06/10/11 22:28, David Nickerson wrote:
> The idea of an archive file works well when you have a single
> transferable set of data that gets shipped from user to user with no
> need to track or merge changes from different users (i.e., one
> developer many users). DOCX archives work because Word provides very
> sophisticated edit tracking and merging functionality - is this
> functionality something that SED-ML tools want to develop and support?
>
> An alternative, that we have been using successfully in the Physiome
> Model Repository, is the idea of using mercurial repositories as a
> "workspace" in which you contain all data related to a piece of work.
> This piece of work could be anything from a complete model with
> simulation descriptions and outputs, experimental data used to fit the
> model, etc. to a single sub-component of a generic model constituent
> designed as part of a component library. While we use mercurial, any
> versioning system could be used instead. The advantage of this
> approach is that the user is able to embed repositories within
> repositories (subrepos in mercurial, submodules in git, externals in
> svn) and manage the versioning of embedded repositories such that the
> user can choose whether to track latest changes or fix on a specific
> version that is know to be "correct".
>
> We find that this allows modellers to easily[1] ship their work
> between each other while maintaining a decent provenance record of the
> developments. While it is not always going to be optimal to include
> all types of data in a single type of repository, mercurial at least
> (and probably the other systems), allows embedding different types of
> repository within a repository. We're starting now to think about how
> this would work when you start wanting to include datasets that are
> huge (e.g., large image collections used to fit patient specific
> cardiac models, or the simulation results of several million cell
> models in a stomach). But at least in terms of the work currently
> being done with CellML models, mercurial  repositories are proving
> sufficient.
>
> When a user simply wants to grab a particular piece of work from the
> repository for their own use, the website (and soon webservices)
> provide a way to download a static archive containing the contents of
> the workspace. Again, we're still trying to work out what to do in the
> case where an workspace links to datasets that can easily be multiple
> gigabytes in size. And we are certainly open to any help on this
> issue.
>
> Cheers,
> David.
>
> [1] There is currently a lack of tool support for this, so users
> pretty much have to manage their mercurial workspaces using mercurial
> directly. This is not always ideal and is also something we are
> thinking about and working on to some degree for certain applications.
>
> On Fri, Oct 7, 2011 at 9:41 AM, Richard Adams<ric...@ed...>  wrote:
>>
>> For those unfamiliar with the concept of the SEDX archive format: It's a
>> zipped archive of a SED-ML file and>=1 model file. Would an equally simple
>> solution be acceptable to incorporate other data files as well?
>> For example, a zip with subfolders
>> models/
>> simulations/
>> data/
>> diagrams/
>> results/
>> Each resource could refer to other resources via relative URIs. E.g., in the
>> scenario above, a SedML file would refer to ../models/model.xml. Of course,
>> within an archive, external public resources can still be referred to if
>> need be - for example, if a dataset is very large.
>> In this approach all the folders are 'source' folders (i.e., inputs to
>> computational tasks) except for results/, which could be used to contain the
>> results of an experiment if need be.
>> In SBSI we have the concept of a 'project' for modelling resources, with a
>> defined folder structure largely similar to that proposed above, to enable
>> tooling to reliably link and detect  resource files, and allow export of
>> zipped archives for exchange between collaborators.  I'm not especially
>> advocating its exact usage but just to say it's worked well for us.
>> Best wishes
>> Richard
>> On 6 Oct 2011, at 17:05, Nicolas Le Novère wrote:
>>
>> It seems to me that the need of an archive is pervasive in all the M&S
>> efforts. We want the models and the simulation descriptions together, but
>> also the simulation results or the data-sets used to complement or fit the
>> models, and of course the SBGN-ML files.
>>
>> And we are going in the same direction in DDMoRe, where the need is clearer
>> since the mathematics only represent part of the "model".
>>
>> After all the  it is just like ODF or DOCX.
>>
>> I therefore propose to move that discussion to combine-discuss.
>>
>> On 06/10/11 16:43, Herbert Sauro wrote:
>>
>> Thank you Richard, that is exactly what I wanted to know, will read appendix
>> D.  The zip file idea has been floated around for a while (since 2004 I
>> think by Cliff Shaffer in particular) but in the past there was never
>> anything to zip with the sbml model until the advent of sedml.
>>
>> Herbert
>>
>> Sent from my iPad, hence excuse my typos.
>>
>> ------------------------------------------------------------------------------
>>
>> All the data continuously generated in your IT infrastructure contains a
>>
>> definitive record of customers, application performance, security
>>
>> threats, fraudulent activity and more. Splunk takes this data and makes
>>
>> sense of it. Business sense. IT sense. Common sense.
>>
>> http://p.sf.net/sfu/splunk-d2dcopy1
>>
>> _______________________________________________
>>
>> SED-ML-discuss mailing list
>>
>> SED...@li...
>>
>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>>
>>
>> --
>> Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
>> Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
>> Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
>> http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/
>>
>>
>> ------------------------------------------------------------------------------
>> All the data continuously generated in your IT infrastructure contains a
>> definitive record of customers, application performance, security
>> threats, fraudulent activity and more. Splunk takes this data and makes
>> sense of it. Business sense. IT sense. Common sense.
>> http://p.sf.net/sfu/splunk-d2dcopy1
>> _______________________________________________
>> SED-ML-discuss mailing list
>> SED...@li...
>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>>
>>
>> Dr Richard Adams
>> Software Development Team Leader,
>> Centre For Systems Biology Edinburgh
>> University of Edinburgh
>> Tel: 0131 651 9019
>> email : ric...@ed...
>> Web: http://csbe.bio.ed.ac.uk/adams.php
>>
>>
>>
>>
>>
>> The University of Edinburgh is a charitable body, registered in
>> Scotland, with registration number SC005336.
>>
>> ------------------------------------------------------------------------------
>> All the data continuously generated in your IT infrastructure contains a
>> definitive record of customers, application performance, security
>> threats, fraudulent activity and more. Splunk takes this data and makes
>> sense of it. Business sense. IT sense. Common sense.
>> http://p.sf.net/sfu/splunk-d2dcopy1
>> _______________________________________________
>> SED-ML-discuss mailing list
>> SED...@li...
>> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>>
>>
>
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure contains a
> definitive record of customers, application performance, security
> threats, fraudulent activity and more. Splunk takes this data and makes
> sense of it. Business sense. IT sense. Common sense.
> http://p.sf.net/sfu/splunk-d2dcopy1
> _______________________________________________
> SED-ML-discuss mailing list
> SED...@li...
> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss


-- 
Nicolas LE NOVERE, Computational Systems Neurobiology, EMBL-EBI, WTGC,
Hinxton CB101SD UK, Mob:+447833147074, Tel:+441223494521 Fax:468,
Skype:n.lenovere, AIM:nlenovere, twitter:@lenovere
http://www.ebi.ac.uk/~lenov/, http://www.ebi.ac.uk/compneur/