Re: [Miase-discuss] MIASE paper

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

> 2) "Whether or not the simulation result matches reality and whether or not
> an experiment can be conducted on a certain model will neither be described
> nor tested."
> You argued that a description is tied to a certain model or number of models
> which makes the comment on scope irrelevant, i.e. whether an experiment can
> be run on the model. But I think what is meant here is to point out that
> there is no validation of whether or not it does make sense to run the
> simulation on the referenced model.

ahhhh...ok, I get that now. Perhaps the "can be conducted on a certain
model" can be changed a bit to make this clearer. Maybe "should be
conducted" ?

> 3) Rules
> 1.B. "If the model is not encoded in a standard format, then the model code
> must be made available to the user."
> You said that: "having a model encoded in a standard format doesn't imply
> that it is freely available. Perhaps might be better to state that model
> descriptions must be MIRIAM compliant?"
> You are right in the sense that there are cases where models are encoded in
> a standard format, but they are still not available... But I think the rule
> was not addressing the distribution policy...
> I am wondering if not the statement in 1. "...All models used in the
> experiment must be named and contain a reference to a model source." might
> be sufficient?

agreed, that should be enough.

> Maybe it would make sense to change it to "a reference to a freely available
> model source". (If we want to force that at all)
> I actually think that people who do not want to share their models might not
> want to share their simulation experiments neither ;-)

thats true, but I guess we want to address the case where people want
to share their work with models hard-coded into proprietary software.
I like the comment (can't recall where it was) that things would just
be better with freely available models in standard formats, but we're
not imposing such in order to be MIASE compliant.

> One more thing here, we should not enforce the use of MIRIAM. We should only
> recommend people to refer to MIRIAM compliant models, if possible.

yeah, I was really just thinking that since MIRIAM addresses
availability, licensing, etc so we can leave it up to MIRIAM and not
need to address it in MIASE. Is it really that bad to enforce the use
of MIRIAM compliant model descriptions?

> 4) The Repressilator example
> You suggested to do "a cell physiology type example as something more
> familiar to the CellML community. Perhaps something along the lines of
> calculating APDs in a parameter scan?"
> The more example we have, the better. And the more diverse the even more
> better.
> So, if you volunteer to set up the example :-) ... I would not say no :-)
> (BTW, Viji is currently doing an example that includes post-processing of
> the output. )

cool. I'm kind of hoping someone might volunteer :) I think the
current example plus Viji's one would be sufficient, but it would be
nice to have more. If I get a chance I'll give it a try.

> 5) Discussion
> By "The sole annotation of models is not sufficient to promote reuse of
> existing biological knowledge." I meant "the annotation of the models alone
> is not sufficient, you also need to information on the simulation as well".
> Maybe I could actually change it to "The annotation of \models\ only"...

sounds good.


Thanks,
David.