My guess is to keep it simple: 1) don't prescribe how simulation software is to calculate this - i.e. don't prescribe a particular mathematical formulation for representing time delays 2) calculate the variable(s) representing the output of delayed function separately - e.g. in a separate component or math element - i.e. don't promote variables to being able to be functions also
I think it is necessarily going to require: 1) math elements that point to user defined code e.g. in the simplest case the code will buffer of a particular variable and perhaps create a local approximation curve around the time required and return a result. 2) referencing math elements from other components e.g. to calculate f(y-T) you need to at least have a reference to the math of y in the case that someone has an analytical method to apply instead of some algorithm over a history of values On 13/02/2008, at 12:08 PM, Catherine Lloyd wrote: > Dear All > > The issue of how a CellML model should handle time delays has been > discussed previously by the Auckland CellML group, and at the time > there > was a tracker item filed on this subject: > > https://tracker.physiomeproject.org/show_bug.cgi?id=70 > > The topic has been recently raised again (see the email discussion > below), and it was discussed at today's CellML meeting (13th Feb, > 2008). > > We'd like to open up this discussion to the CellML community to see if > anyone has any thoughts, ideas, or preferences as to how time delays > should be a) expressed in a CellML model and b) handled by the > software > tools. > > Thank you in advance for your feedback. > > Best wishes > Catherine > >> Hi Rui, >> >> If you are already familiar with Virtual Cell, I'd stick to that. The >> initial version of SBML was largely based on the internal VCML >> used by >> the Virtual Cell, and I suspect the VCell developers remain involved >> in recent developments within the SBML community. >> >> In the past, VCell was capable of importing CellML models but I seem >> to recall that there might be issues with that in recent versions. If >> it is still possible, you should be able to import the CellML version >> of the model and then touch it up from there (add in the time delay). >> You should then be able to export the SBML version for use in other >> tools if required. >> >> CellML2SBML is a collection of XSL stylesheets which are capable of >> processing most models from the repository. As Catherine mentioned, >> you'll lose the metadata from the CellML model in the resulting SBML >> model, but the math should largely be intact. From my understanding, >> the CellML2SBML stylesheets are mainly targeted at CellML models >> which >> use the now deprecated reaction element, so if the model you are >> interested in uses reactions then you'll get the best SBML version. >> Otherwise I suspect you'll end up with a generic SBML encoding of the >> mathematics. >> >> Oh, and I'm pretty sure VCell will allow you to specify time delays >> although I can't remember the specific syntax or interface options >> and >> you can see https://tracker.physiomeproject.org/show_bug.cgi?id=70 >> for >> a bit of discussion on this issue within CellML - feel free to add to >> that discussion! >> >> Not sure if I can offer much more help than that but let me know if >> you have any more questions... >> >> >> David. >> >> >> Catherine Lloyd wrote: >> Dear Rui >> >> You are welcome! >> >> Hmmm, much as I would hate to lose a potential CellML user to SBML, I >> can see the advantage for you if indeed, SBML can handle time delays. >> Although I'm not familiar with the structure of SBML, my >> understanding >> is that the cellml2sbml converter is not perfect - if nothing >> else, it >> strips all the metadata out of a CellML model. But you might be >> able to >> use it as a foundation which you can then edit by hand. >> >> I have cc'd David Nickerson in on this email - if anyone deserves the >> title of "CellML expert" it is David!... David do you have any idea >> which SBML tool might be best to use in this case? >> >> I've also cc'd Poul Nielsen and Peter Hunter who lead the CellML >> project. I'm sure they will be interested in this case and they may >> have an idea how long it might take to develop the CellML tools such >> that they are capable of handling time delays. >> >> I'm sorry I can't help you more myself, but I imagine we can work >> together to get this up and running for you. >> >> Best wishes >> Catherine >> >> >> >> Rui Zhu wrote: >>> Dear Catherine, >>> >>> Thank you so much for your reply! It is very helpful! Learning >>> CellML >>> modeling from an expert is a really delightful experience. >>> >>> I am having another thought. I searched online, and it says SBML >>> level2 can deal with the time delay term, and there is a software >>> (cellml2sbml) to convert CellML to SBML. Do you think it is a good >>> idea to use SBML to solve the problem? And from SBML website, there >>> are a lot of tools that can run SBML model, like virtual cell. I am >>> wondering if you have any idea which tool can handle the time delay >>> term. Thanks! >>> >>> Your suggestion will be appreaciated! >>> >>> Many thanks, >>> >>> Rui ZHU >>> >>> >>> ----- Original Message ----- >>> From: Catherine Lloyd <[EMAIL PROTECTED]> >>> Date: Sunday, February 10, 2008 11:19 am >>> Subject: Re: time delay term in CellML model >>> To: Rui Zhu <[EMAIL PROTECTED]> >>> Cc: Poul Nielsen <[EMAIL PROTECTED]>, Peter Hunter >>> <[EMAIL PROTECTED]> >>> >>> >>>> Dear Rui >>>> >>>> Thank you for your interest in CellML. I am familiar with the >>>> model >>>> you >>>> are interested in, and I know the original model author (Michael >>>> Zagar)helped me to translate this model into CellML. I'm sorry >>>> if our >>>> curation status confused you - it does indeed state that the CellML >>>> model recreates the results of the published model - but it should >>>> statethat the models almost match - with the exception of the time >>>> delay(I'll change the text now). >>>> Unfortunately this is a current limitation of CellML. While it is >>>> possible to express a time delay in the CellML language, the tools >>>> whichrun CellML models (PCEnv and COR) are unable to handle time >>>> delays at >>>> present. This is something which we expect to be a feature in >>>> future >>>> releases of the software. >>>> >>>> Please let me know if you have any further questions and I will >>>> try to >>>> answer them for you. >>>> >>>> Best wishes >>>> Catherine >>>> >>>> >>>> >>>> Rui Zhu wrote: >>>> >>>>> Dear Catherine, >>>>> >>>>> This is Rui ZHU, a Biomedical Engineering Ph.D student in >>>> University of Southern California. I am very interested in using >>>> CellML to do modeling. Recently, I saw the CellML model you made >>>> based on the paper "A Delayed Nonlinear PBPK Model for Genistein >>>> Dosimetry in Rats" on the CellML website. >>>> http://www.cellml.org/models/zager_schlosser_tran_2007_version03 >>>>> And I tried to use that model to do simulation using PCEnv to re- >>>>> >>>> generate the results in the original paper. However, I found the >>>> result was different from the one the in original paper, and the >>>> reason it that, in the original paper, there is a time delay >>>> term in >>>> the differential equation d( Acon_b )/d(time) = kbile * H_Ccon_l >>>> * Ccon_l ( t - d(t, C_gen_l) ) * Vl, but in the CellML model the >>>> differential equation is without the time delay term d( Acon_b >>>> )/d(time) = kbile * H_Ccon_l * Ccon_l * Vl. I was trying to >>>> add a time delay term in the CellML model, but I don't know how. If >>>> you have any idea of how to do it, would you please let me know. >>>> Thank you! >>>>> Your kind help is really appreciated! Look forward to your reply! >>>>> >>>>> Rui ZHU >>>>> >>>>> >>>> > > _______________________________________________ > cellml-discussion mailing list > [email protected] > http://www.cellml.org/mailman/listinfo/cellml-discussion _______________________________________________ cellml-discussion mailing list [email protected] http://www.cellml.org/mailman/listinfo/cellml-discussion
