asplund-samuelsson / thermosampler Goto Github PK

All transporter reactions are currently modeled as passive with a standard delta G of 0 kJ/mol. Some/all of these reactions may have other mechanisms, such as transport powered by symporting, antiporting, or ATP. This question is particularly important for the SPOB acetate export. However, active acetate export or uptake could not be confirmed in literature searches.

Explain how to use the config file with plot_samples.R. It has three tab-delimited columns (Id, Name, and Type), and is used to order, filter, and rename the Metabolite, Reaction, and Group variables to be plotted.

At the moment it is possible to limit the total concentration of all metabolites during the random walk by supplying a single value. It would be good with the option to have multiple custom groups of metabolites that each has its own maximum total concentration. One application is to simulate multiple cells with separate metabolite pools.

A suggested format for defining these groups would be a table like this, where the columns represent Metabolite, Group, and Sum:

NADH     1     0.01
NAD      1     0.01
H2O      2     1
ATP      3     0.05
ADP      3     0.05
AMP      3     0.05

Metabolites not defined in the table could be treated under the default/system-wide metabolite concentration sum.

Currently acetate is transformed into acetyl-CoA in a single step in the acetoclastic methanogen of the syntrophic model (reaction ac_pct). This should perhaps be modeled more accurately as split between acetate kinase and phosphotransacetylase reactions.

Shalvarjian and Nayak (2021) point out that the ACDS reaction generates a CO group that is oxidized to CO2. Perhaps the ac_acs reaction in the syntrophic model should be updated.

Some metabolite concentrations from the hit-and-run sampling appear to be infeasible when loaded into other scripts/programs. They may give certain reactions a very small positive delta G, e.g. +0.01 kJ/mol or +0.024 kJ/mol. This suggests a rounding error in the metabolite concentrations. To alleviate this problem, it is suggested to set a concentration "precision" variable that determines how many significant digits should be used in the metabolite concentrations (natural log scale) both during calculation and in the saved output. Note that this new parameter should not be confused with the "precision" variable that determines how close the random walk can get to the edge of the solution space. A bonus should be smaller file sizes if the number of significant digits is reduced.

Larger models and many random walk steps yield very large output files. It would be good to have the option to only save every nth feasible metabolite concentration set.

Currently many metabolite concentration ranges are defined by default values (basically unconstrained, within reason) or by values taken from E. coli. Experimental data more closely describing the SPOB or syntrophic model conditions may be available.