[hoffmann]

suppose we could reconstruct a metabolic network using statistical tests from some sort of high throughput data.  not exactly the typical metabolic pathways in biochemistry books like tca, glycolysis, etc. but actual metabolite-metabolite interactions based on their in vivo concentrations.

1. how do we measure the in vivo concentrations of metabolites?
2. what use would this kind of information have in biology?  are metabolite-metabolite interactions in a network meaningful?
3. instead, what if we were able to predict metabolic reactions from a high throughput data set.  would this be more useful than the former?
4. do these kinds of algorithms exist to reconstruct metabolic networks?  i've done a literature search in systems biology journals but no matches for this kind of method.

thx!

[lancenti]

1. The problem with this is that depending on the type of reaction, and since it keeps going, we're going to not be able to 'freeze' the reaction to go measure. What we can do is to saturate the system at points to see the maximum output, and probably step it down from there slowly and use the data to derive the nature of the system (where slow-step happens, if there's a steady-state equilibrium, etc.) Of course, this would probably kill the subject. We could, of course, devise a scale model of a human body but I'm not sure how well that will work.

2. Metabolite-Metabolite interactions are meaningful, I believe, because if two metabolites are totally inert as individuals but become toxic/have some side-effect it may become a very good drug or a pair of drugs not to be administered together.

3. I think you can only get kinetic data from the former type.

4. I have no idea.