tentusscher_2004

Penny has curated this model nicely. 3 variants: 0.) M-cell 1.) epicardial 2.) endocardial

versions 04 and 05 are duplicates of v3v1, made for Peter's sake. v3v1, v4 and v5 all have PCEnv session files. v5 has a souped up session complete with interactive SVG diagram.

Version 5 almost produces completely accurate replication of results, except for the L-type Ca++ channel current, which is slightly off for some reason.

Brian Carlson of the Medical University of Wisconsin sent Peter Hunter an email in early March 2008 describing his experience with using this model in JSim. He acknowledged that the model produced the correct results but noted that it was mathematically different from the original publication. He suggested that we need a much better version history than we currently have. The problem here is that Penny Noble curated the model and sent it to me, but she did not make notes on what she changed at the time to get the model working. The following is a transcript of the relevant communications - at the top is Penny's reply to my request for some notes on what she did with the TT04 model, and at the bottom is Brian Carlson's original email to Peter. Kirsten ten Tusscher's C code that Penny refers to can be found at http://www-binf.bio.uu.nl/khwjtuss/SourceCodes/

"Dear All,

Just to confirm I have checked what I did with the TT model and as I expected I worked and curated it straight from Kirsten's code, which I downloaded from her website (see attached) and hence do not have notes on changes between the paper and cellml code, which should be an exact match to Kirsten's C code. The values and units for R and F in the paper are indeed different to those in the code.

Also Kirsten's code simplifies the equation for IK1 by removing the Ko dependence term, but we added this back in as it could be important for simulations where Ko is not 5.4mM.

The other issue this end was that this model was unit curated by someone other than myself (Martin Fink), as he was working on using the model in any case and wanted to try out Alan's then new ability of COR to unite curate. I believe he made no notes on the process of that curation as he was simply focused on getting the model warning free, but I do recall him telling me that the units he had to use for ICaL were odd!

Best wishes, Penny

-----Original Message----- From: penny.noble@dpag.ox.ac.uk mailto:penny.noble@dpag.ox.ac.uk Sent: 03 March 2008 11:05 To: Peter Hunter Cc: James Lawson; Catherine Lloyd; Penny Noble; Poul Nielsen; Randall Britten Subject: Re: ten Tusscher model and Model curation

Dear All,

James has also emailed me asking if I have notes on the changes made to the models during the curation process. I have to confess in many cases what I did was to either email authors and inplement any corrections they told me about, or to check against the code (if available) and curate them in relation to code rather than the papers; both of which means that I don't have any specific notes for the changes for many of the models from what was in the paper.

In relation to the ten Tusscher model I curated it against Kirsten's code on her website and hence it ought to match that. In addition when we got down to the unit curation process over here, using COR, we found some units were incorrect and hence changes were made to correct those, including values for R and F. The model you have ought to run in COR without any errors.

I am not in the office today as I'm off sick but can check when I go back in what other information I have.

Apologies as indeed I can see more explicit notes would have been helpful.

Best wishes, Penny

Peter Hunter wrote: > > Dear Brian, > > > > Thanks for your detailed and useful comments. I've CC'd James Lawson, > > Catherine Lloyd (both > > Auckland) & Penny Noble (Oxford) who are all curating CellML models. I > > think the lesson for us from > > this is that we need to be more explicit about the changes from the > > original made when curating a > > model. I've also CC'd Randall to raise this at the CellML mtg next > > Wednesday. > > > > Cheers, > > Peter > > > > Carlson, Brian E. wrote: >> >> Dear Jim and Peter, >> >> >> >> >> >> >> >> I have just converted the ten Tusscher et al. human ventricular myocyte >> >> model from CellML to JSim and I wanted to share with both of you my >> >> experience while my impressions are still fresh. >> >> >> >> >> >> >> >> From the ten Tusscher model page on CellML >> >> >> >> >> >> >> >> http://www.cellml.org/models/tentusscher_noble_noble_panfilov_2004_version05 >> >> >> >> >> >> >> >> I downloaded both the PCEnv and JSim versions and tried to run each of >> >> them. The PCEnv version replicated the results in the paper >> >> successfully whereas the JSim version successfully compiled but did not >> >> produce the responses in the paper. Discovering the reasons for this >> >> discrepancy took the better part of a week and I will summarize what the >> >> major issues were in the reconciliation of these models: >> >> >> >> >> >> >> >> 1) PCEnv uses version 5 of the model and JSim uses version 1. >> >> >> >> 2) PCEnv version 5 used equations which had been changed from the >> >> paper but were not overtly documented. >> >> >> >> 3) JSim turned unit conversion off. >> >> >> >> 4) JSim made a straight conversion of units to the CellML format >> >> which had to be undone to confirm that the units were correct. >> >> >> >> 5) JSim was not able to import the initial conditions. >> >> >> >> 6) Some units in the PCEnv version (and CellML version 5) have >> >> been altered to generate the results in the paper (e.g R=8314.472 >> >> J/(K*mole) and F=96485.314 C/mmole). It all cancels out in the end but >> >> looks problematic. >> >> >> >> >> >> >> >> The PCEnv simulation environment was easy to learn even though, of >> >> course, I am much more familiar with JSim. I will continue to use PCEnv >> >> even though for the current application we have at MCW the convenience >> >> of changing parameters over a range of values in JSim (or Matlab) will >> >> be key. I have put up the JSim version on the UW Physiome site at: >> >> >> >> >> >> >> >> http://www.physiome.org/model/doku.php?id=Cell_Physiology:Action_potential:t en_Tusscher_Noble_Noble_Panfilov_2004:model_index >> >> >> >> >> >> >> >> In the JSim source code you will see a much more detailed account of >> >> what was changed for this specific model and why. >> >> >> >> >> >> >> >> I do have to admit that replicating a model in another programming >> >> language which would normally take many weeks to accomplish took just >> >> one because of the tools available from NSR and CellML. Thanks. >> >> >> >> >> >> >> >> Sincerely, >> >> >> >> >> >> >> >> Brian >> >> >> >> >> >> >> >> **************************************************************** >> >> >> >> Brian Carlson >> >> >> >> Research Scientist >> >> >> >> Biotechnology and Bioengineering Center >> >> >> >> Medical College of Wisconsin >> >> >> >> 8701 Watertown Plank Road >> >> >> >> Milwaukee, Wisconsin 53226 >> >> >> >> >> >> >> >> phone: (414) 955-4822 >> >> >> >> cell: (206) 437-1374 >> >> >> >> becarlson@mcw.edu <mailto:becarlson@mcw.edu> >> >> >> >> >> >> >> >> **************************************************************** >> >> >> >> >> >> > >"