Re: [SED-ML-discuss] Fwd: [sed-ml-discuss] Figures from HARMONY, plus extras

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Hi all,

In my opinion the repeated task is already handling the multiple stochastic
runs.
The only issue is that we have to clearly define in the spec what is the
expected output of repeated task and nested repeated task in form of data
matrices which come out of them.

How I understand the repeated tasks, and implemented them, is that with
every nesting an additional matrix dimension is created.
* Simple task of Time course simulation has result M(Nt, 1), i.e. matrix
dimension of N time points times 1
* Simple task of One step and steady state has M(1,1)
A repeated task r1 has
* M(Nt, Nr1) for time course
* M(1, Nr1) for steady state and one step
A repeated task r2 of repeated task r1 (like scanning parameters over
stochastic repeats, or parameters against parameters, i.e. scanning a
parameter grid) results in
* M(Nt, Nr1, Nr2) timecourse
* M(1, Nr1, Nr2) one step and steady steady state
and so on, i.e. with every nesting an additional dimension is added to the
result matrix.

What we need is mainly calculation of functions on dimensions of this
matrix, i.e.
mean(data, dimension) or std(data, dimension), i.e. the calculation on
certain dimension of the resulting data matrices. In the following indexing
starting from 0, i.e.
functions on dimension 1 are integrating repeated task 1, functions on
dimension 2 repeated task 2 and so on.
The simple stochastic moments can than be calculated as

mean(M(Nt, Nr1), 1) and std(M(Nt, Nr1), i.e. the functions are evaluated
over the repeats
If one wants a parameter scan over repeated tasks one would first scan the
repeats for a given parameter (r1) and in the inner repeated task (r2) vary
the parameters.
Afterwards one calculates
mean(M(Nt, Nr1, Nr2, 1) and std(M(Nt, Nr1, Nr2, 1) which brings down the
dimension of the result matrix by one.

That is why it is imporant to define what the result dimensions the
different tasks can provide. For instance Jacobian tasks would produce
M(Ns, Nv), with repeated jacobian adding more dimensions.
In my opinion we have to write down once for every task what the result
dimensions are, and how repeated tasks extend the dimensions.
Than we have to allow function calculations on certain dimensions of the
result Matrices.

M

On Wed, Jul 5, 2017 at 5:58 AM, David Nickerson <dav...@gm...>
wrote:

> Hi all,
>
> To keep up with this exciting discussion, don't forget to sign up to
> the new google group over at
> https://groups.google.com/d/forum/sed-ml-discuss.
>
> Cheers,
> David.
>
>
>
> ---------- Forwarded message ----------
> From: Chris Myers <my...@ec...>
> Date: 4 July 2017 at 09:17
> Subject: Re: [sed-ml-discuss] [SED-ML-discuss] Figures from HARMONY, plus
> extras
> To: Lucian Smith <luc...@gm...>
> Cc: sed...@go...
>
>
> No, slice will not work.  This is not for reading a data file
> externally, this is for specifying a data generator.  Here is how I do
> this now:
>
>     <dataGenerator id="IPTG_Environment_mean_dg" name="IPTG">
>       <annotation>
>         <dataSet xmlns="http://www.async.ece.utah.edu/iBioSim"
> dataset="mean" />
>       </annotation>
>       <listOfVariables>
>         <variable id="IPTG_Environment_mean_var" name=""
> taskReference="Environment"
> target="/sbml:sbml/sbml:model/sbml:listOfSpecies/sbml:species[@id='IPTG']"
> />
>       </listOfVariables>
>       <math:math>
>         <math:ci>IPTG_Environment_mean_var</math:ci>
>       </math:math>
>     </dataGenerator>
>     <dataGenerator id="IPTG_Environment_stddev_dg" name="IPTG">
>       <annotation>
>         <dataSet xmlns="http://www.async.ece.utah.edu/iBioSim"
> dataset="stddev" />
>       </annotation>
>       <listOfVariables>
>         <variable id="IPTG_Environment_stddev_var" name=""
> taskReference="Environment"
> target="/sbml:sbml/sbml:model/sbml:listOfSpecies/sbml:species[@id='IPTG']"
> />
>       </listOfVariables>
>       <math:math>
>         <math:ci>IPTG_Environment_stddev_var</math:ci>
>       </math:math>
>     </dataGenerator>
>     <dataGenerator id="IPTG_Environment_1_dg" name="IPTG">
>       <annotation>
>         <dataSet xmlns="http://www.async.ece.utah.edu/iBioSim"
> dataset="1" />
>       </annotation>
>       <listOfVariables>
>         <variable id="IPTG_Environment_1_var" name=""
> taskReference="Environment"
> target="/sbml:sbml/sbml:model/sbml:listOfSpecies/sbml:species[@id='IPTG']"
> />
>       </listOfVariables>
>       <math:math>
>         <math:ci>IPTG_Environment_1_var</math:ci>
>       </math:math>
>     </dataGenerator>
>
> What I would prefer to do instead is be allowed to treat the variable
> id in a stochastic run as an array, and in the math be able to do
> things like:
>
> mean(IPTG)
> stddev(IPTG)
> IPTG[1]
>
> This would be much cleaner.
>
> Chris
>
> On Jul 4, 2017, at 10:12 AM, Lucian Smith <luc...@gm...>
> wrote:
>
> No, we didn't discuss either of those proposals explicitly.  For the
> second, I was under the impression that the 'Slice' construct could
> get you what you want, but I don't know this for sure.  Frank, would
> that work? If not, I bet an update to 'slice' would make it work.
>
> -Lucian
>
> On Mon, Jul 3, 2017 at 9:31 PM, Chris Myers <my...@ec...> wrote:
> >
> > Hi Lucian,
> >
> > Thanks for creating these.  I like the updates for plots and curves.
> However, I do not see any updates for number of runs for a Task and a means
> to access specific run data (or statistics) by a data generator.  I was
> pretty sure we agreed on that at COMBINE 2016.  Did these get discussed at
> Harmony?
> >
> > Chris
> >
> > > On Jul 3, 2017, at 9:05 PM, David Nickerson <dav...@gm...>
> wrote:
> > >
> > > ---------- Forwarded message ----------
> > > From: Lucian Smith <luc...@gm...>
> > > Date: 3 July 2017 at 13:53
> > > Subject: [SED-ML-discuss] Figures from HARMONY, plus extras
> > > To: sed...@li...
> > >
> > >
> > > As you may have heard by now, HARMONY 2017 was a very productive
> > > conference for the SED-ML community.  Many proposals that have been
> > > considered and thought about for a long time were finally made into
> > > concrete UML diagrams, and work has begun on writing up a proposed
> > > specification describing those changes.
> > >
> > > I was in charge of creating the UML diagrams, and while I contributed
> > > to their content, it was by no means a solo effort.  Over the weekend,
> > > I've worked on updating them a little more, and have thrown in a
> > > couple other proposed changes while I was at it.  Some of the proposed
> > > changes would be fine in a new l1v4 document, while others might be
> > > best suited to an l2v1; no decisions have been made about either
> > > possibility.
> > >
> > > It will probably make the most sense to talk about each set of
> > > proposed changes separately, to make sure everything is properly
> > > discussed, but if you're interested in perusing them in the meantime,
> > > the whole folder is at:
> > >
> > > https://drive.google.com/drive/folders/0B71r5kLs3FlibnZMUEpIU1RKR1k
> > >
> > > In particular, take a look at the legend, to make sure you know what
> > > you're looking at:
> > >
> > > https://docs.google.com/drawings/d/1BEzZfK0MyA8qyplkrIvUkEpGeDr1c
> 8Ji1LPx_6pz9eM/edit
> > >
> > > In general, a red color means it's a newly-proposed change from l1v2.
> > >
> > > Also note that 'id' and 'name' have been proposed to be moved to
> > > SEDBase, which explains why they are missing from some other figures.
> > >
> > > If anything is incorrect or confusing, let me know, and I will make
> > > updates!  And thanks in particular to Frank, Herbert, Matthias, and
> > > David for their work and contributions during the conference.
> > >
> > > -Lucian Smith
> > >
> > > ------------------------------------------------------------
> ------------------
> > > Check out the vibrant tech community on one of the world's most
> > > engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> > > _______________________________________________
> > > SED-ML-discuss mailing list
> > > SED...@li...
> > > https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
> > >
> > > --
> > > You received this message because you are subscribed to the Google
> Groups "sed-ml-discuss" group.
> > > To unsubscribe from this group and stop receiving emails from it, send
> an email to sed...@go....
> > > To post to this group, send an email to sed-ml-discuss@googlegroups.
> com.
> > > To view this discussion on the web, visit https://groups.google.com/d/
> msgid/sed-ml-discuss/CADizRjgi%2BbvOOQvxP7n%3DA0duBAOfLO2L4No7%
> 2BGJbtAfaB3-JFQ%40mail.gmail.com.
> > > For more options, visit https://groups.google.com/d/optout.
> >
>
>
> --
> You received this message because you are subscribed to the Google
> Groups "sed-ml-discuss" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to sed...@go....
> To post to this group, send email to sed...@go....
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/sed-ml-discuss/B65C820D-
> 6BD0-429B-BE2A-FCCC844389B8%40ece.utah.edu.
>
> For more options, visit https://groups.google.com/d/optout.
>
>
> --
>
>
> David Nickerson
> about.me/david.nickerson
>
> ------------------------------------------------------------
> ------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> SED-ML-discuss mailing list
> SED...@li...
> https://lists.sourceforge.net/lists/listinfo/sed-ml-discuss
>

-- 
Dr. Matthias König
Junior Group Leader LiSyM - Systems Medicine of the Liver
Humboldt-University Berlin, Institute of Biology, Institute for Theoretical
Biology
  https://www.livermetabolism.com
kon...@go...
Tel: +49 30 20938450
Tel: +49 176 81168480