Jim Bassingtwaighte (March 7, 2010):
We are brainstorming with the overall goal of determining community/group efforts we can focus on to forward research in muscle physiology.
I have a variant on your last suggestion. Given that we all need a big model of intermediary metabolism (Mito energetics, TCA, NAD/NADH/ phosphoenergetics, ion cycling, pentose shunt, insulin recpetor regulation, and such) we might take individual pieces of these and plan to put them together as a group.
We would thn have a standard semantic framework, a standard description of each of the reactions (bidirectional, thermodynamically constrained), parameter sets for each module with references etc.
Modular design would be critical. I would suggest doing it on two platforms, Matlab and JSim, so that everyone will be happy that they have something which is easy to use and portable.
Archival format isn't so easy since some components or integrated systems will need PDEs. One dimensional ones can be kept in CellML I think, but the vehicle for computing CellML based archival form is currently only JSim and that relationship is not thoroughly tested. We do not yet have a JSim to CellML converter either, though JSim is a lovely simple language in which to formulate the models. Just write the differential equations, ODEs or PDEs.
We would have a common database somewhere, presumably with you for each enzyme. We could look after transporters; Somebody else fo channels, or whatever.
Osmotic events would soon need to be included.
Different groups might like to focus particular cases on different cell types. Or different species, The parameters will differ in any case, and sometimes the model structures.
Obviously we need a group consensus on a lot of little details before thsi could work. But the idea is to be using the pieces as we construct the whole, so that the contributor to the whole is not just supplying the info on one enzyme after another, but building his own thing and making it contribute to a larger whole.
Peter Hunter (March 7, 2010):
I'll talk to Edmund and others tomorrow but I'd say we'd be a starter for this. The CellML framework for handling the modular components is in place and the metadata tools for annotating model components against biological and biophysical ontologies is being put in place with the Ricordo project being led by Bernard de Bono at EBI. There is still quite a bit to do on the parameter databasing side but Jim's proposal would be a great test for all this and the JSim/OpenCell simulation environments.
Dan Beard (March 7, 2010):
These are great ideas and represent the kind of community effort that I hope we can use this as a vehicle for. I do think that we need to have a WG net meeting of the interested parties to shape what we want to/can do. While I agree that we can "take individual pieces", I am sure we also all understand that we need to sort out the multiple levels or organization from the get-go. I think that most of the things we need are mentioned below, but let me take a preliminary crack at some overall organization principles. The main categories of things we need are: thermodynamics, kinetics, databases, and software:
1. Thermodynamics. For metabolism kinetics are highly sensitive to the thermodynamics on which the models are established. In addition, every reaction has a non-integer effective proton stoichiometry that is a fundamentally tied to the thermodynamics. We really can't get the kinetics right until you have the thermodynamics right. We have a good starting point with glycolysis and the TCA cycle and we are currently working on an update with pentose shunt and a few other related reactions. Since the thermodynamic parameters are all interrelated, the database become more accurate as it grows. We have a preliminary pilot web resource for thermodynamic data that you might look at (http://bbc.mcw.edu/Computation/models/thermodata/BioThermoDatabase.htm). We have not yet taken on beta-oxidation because of the sparseness of data. Yet, I think that we will be able to do it--it is mainly just a matter of manpower.
2. Kinetics. Like the case of thermodynamics, there exist loads of relevant kinetics data, largely unanalyzed, and too often misinterpreted. Our approach in terms of enzymes and mitochondrial ion transporters has been to systematically discover and build appropriate models that can explain the available data. On the enzyme side, we have completed a number of these studies: a. citrate synthase (http://www.plosone.org/article/info:doi%2F10.1371%2Fjournal.pone.0000144) b. isocitrate dehydrogenase (http://dx.doi.org/10.1016/j.bbapap.2008.07.001) c. fumarase d. complex-I e. malate dehydrogenase
I list these out to illustrate three of points: First, progress is slow. Two are published, and the remaining three are currently being written up. Second, in every case either the agreed upon mechanism was not able to match the data that the agreed upon mechanism was based upon or there was no agreed upon mechanism. So for the most part we can forget about getting parameters from the literature for these things--we cannopt even get mechanisms from the literature. Thus there is a lot of work to build up these models. Third, the citrate synthase study illustrates a very important link between the thermo and the kinetics. Paper a above assumes Alberty's thermodynamic parameters. Yet our new improved thermodynamic numbers differ substantially. I just recently reparameterized the CTS model based on the improved thermo data and the result is a very different parameter set. Luckily the basic model still works and we only need to make an updated supplement available. An important lesson is that given incorrect thermo data you might be able to find associated (incorrect) kinetic parameters that combined together can explain the in vitro data. But, you should be very cautious to apply the resulting model. I also mention that Ranjan Dash has lead a very productive approach to build up a similar database for mitochondrial ion transporters.
3. Databasing. This effort is very important and needs to be linked to the above two. I would argue that raw data databasing is much more a primary concern than parameter databasing at this stage. (At least for metabolic modeling.) Before we can do too much about databasing the parameters, I want to worry about having the right models (kinetic and thermodynamic) with which the parameters are to be associated. Then we need to agree on data. Often we have to collect new data. (All of these data should be archived.) Then we are in a position to estimate the parameters. And then we can put the numbers in a database. All of this is not to say that we don't need to think about the parameters and get started. Ideally, all of the above is merged into a self-consistent framework.
4. Software. It goes with out saying that we need software to bring all of these things together.
Finally, putting this together I am afraid that I might have gotten off track "broad needs" and outlined my lab's own current research efforts. So please forgive me for that--it is hard to not imagine that everybody wants to do what we want to do :) Considering just the thermodynamics, I originally raised that as something that I am sure the whole community needs. The rest is something that I argue the whole community needs, but some would probably argue with me.
So, final final point...some questions for the whole group:
1. Who else is interested in taking on Jim's proposal? Of those that are interested, what is you take on my plan of attack? Can we narrow it down in a way that makes sense to the most people? (Or do we need a completely different direction?)
2. Of those that are not interested, what do you want to do?