Using Metabolomics or Proteomics data as input in COPASI

[Hikmet Emre Kaya]

Hello everyone,

I have mass spec data for relative metabolite levels across mcf10a and a few cancer cell lines. I'm wondering if it's possible to use them as species input for initial values, as there is no direct correlation between metabolite level and molar concentration (absolute or relative). In short, can one directly use mass-spec metabolomics and/or proteomics data for deterministic simulations in COPASI?

I hope my question is clear.

Many thanks,

Hikmet Emre Kaya

[Pedro Mendes]

Hi Hikmet,

Your question is very pertinent as this is the kind of data that is mostly available for metabolites. However you also point out the problem with most of metabolomics data:

 - metabolite level is not something that is easily related to a molar concentration.

This is not just because these are ratios against a reference, but also because different metabolites are detected with different properties (ionization strength, etc.) So metabolite levels are not easily comparable across metabolites.

An very rigorous interpretation would be that metabolite levels are at most qualitative measurements that give you the indication that metabolite concentrations increased or decreased. Some people will be a bit more adventurous and consider that the metabolite levels indicate also how much they increased or decreased (fold changes, as in RNA measurements).

So, if you have some way of relating the metabolite levels from a metabolomics experiment with some molar measurements (from other experiments perhaps), then you could use the fold changes and multiply them by that base. But I think this is probably going to be hard to do.

Some metabolomics experiments are better, though, as they will use "internal standards" (usually C13 labeled metabolites, or some other stable isotope) and those therefore provide quantitative measurements which should be more easily converted to molar concetrations (one problem that still remains is how much dilution happened between the biological system and the sample injected into the mass spec, but that is the same factor for all metabolites).

In short, I'm afraid that most metabolomics data is not so useful directly to include as initial conditions of a model. Perhaps it is a good guide to the relative magnitudes of each metabolite (eg. one is high, the other needs to be much lower, etc)

I would also be interested in methods that could use these data!

Pedro

[Hikmet Emre Kaya]

Dear Pedro,

Many thanks for your elaborate response. The bottleneck in my opinion (and what drove me to look at mass spec data) is: the available systems models on cancer or other diseases collect and compile metabolite concentrations and parameters values from a myriad of different resources. A study of interest bases their model on a previous computational model, which obtained their values from 10 different studies. It then becomes a problem of consistency and reproducibility, e.g., when you want to add on to it and/or make comparative simulations across healthy and several cell lines of a specific cancer.

As you said, mass spec metabolomic profiling is the only way I have seen numerical values provided across different cell lines within the same study. But how it can be leveraged in chemical reaction networks remains a mystery :(