CUE on different carbon sources

[hgscott]

Daniel wants to make plots of the CUE of Amac on different carbon sources:

To do:

• [x] CUE on glucose/acetate/mixture
• [ ] Curate the list of carbon sources to fit nicely into fewer classes
• [x] Run simulations
• [ ] Plot box plot of CUE for the different classes
• [ ] For the individual classes, plot c fates for the individual c sources
  • [ ] Make sure x axis is readable
  • [ ] Sort by size of bar/CUE
  • [x] Order bars so that CUE is easily determinable
• [x] Get metrics for the individual carbon sources
  • [x] C:N ratio
  • [x] Degree of oxidation (Formula in the Westerhoff paper in the review)
  • [x] Enrypoint into central carbon metabolism (Not doing)

[hgscott]

For glucose, acetate, and a mixture of the two: I started with LB = -10 for each alone, and then for both together. Which generated these plots:

[hgscott]

Now I need to change the bounds to better match what Zac did experimentally.

He always did 12 mmol/L carbon, divided into:

• 12 mmol/L glucose
• 12 mmol/L acetate
• 8 mmol/L glucose & 4 mmol/L acetate
• 4 mmol/L glucose & 8 mmol/L glucose

I think I will try to keep the number of carbon atoms the same as when I did -10 LB for glucose, since that seems to be a commonly used bound in other simulations.

Exp. concentration	Bounds
12 mmol/L glucose	LB_glc = -10
12 mmol/L acetate	LB_ace = -60/2 = -30
8 mmol/L glucose & 4 mmol/L acetate	LB_glc = (2/3)x-60 / 6 = -40 / 6 = -6.666 & LB_ace = (1/3)x-60 / 2 = -20/2 = -10
4 mmol/L glucose & 8 mmol/L glucose	LB_glc = (1/3)x-60 / 6 = -20/6 = -3.333 & LB_ace = (2/3)x-60 / 2 = -40/2 = -20

[hgscott]

That gave me these plots:

[hgscott]

While these aren't necessarily CUE, CUE & GGE are easily interpreted from the bars- and I am saving the specific values in a table.

[hgscott]

For the other carbon sources, I first need to do a quick and dirty check of what it grows on.

The model has 45 exchange reactions for carbon-containing compounds, for the following compounds:

L-Tryptophan_e0 Fe-enterochlin_e0 L-Lysine_e0 L-Isoleucine_e0 Enterobactin_e0 PAN_e0 L-Asparagine_e0 L-Alanine_e0 Folate_e0 Heme_e0 ocdca_e0 Niacin_e0 S-Adenosyl-L-methionine_e0 L-Leucine_e0 L-Arginine_e0 Methanol_e0 Glycerol_e0 Myristic acid_e0 L-Tyrosine_e0 L-Serine_e0 L-Histidine_e0 Acetoacetate_e0 Acetate_e0 L-Cysteine_e0 L-Glutamine_e0 Putrescine_e0 D-Glucose_e0 L-Valine_e0 Formate_e0 Butyrate_e0 Thiamin_e0 Palmitate_e0 L-Proline_e0 Riboflavin_e0 Spermidine_e0 L-Aspartate_e0 BIOT_e0 CO2_e0 Pyridoxal_e0 L-Phenylalanine_e0 Cbl_e0 L-Glutamate_e0 L-Threonine_e0 Glycine_e0 L-Methionine_e0

But we only get growth on

L-Alanine Glycerol L-Serine Acetate D-Glucose L-Aspartate

All of which had pretty different carbon fate distributions:

I'm not sue why acetate's uptake is lower than everything else.

[hgscott]

With this few carbon sources, is breaking into the different classes really necessary?

What would the classes be?

• Amino Acids
  • L-Alanine
  • L-Serine
  • L-Aspartate
• Monosaccharides
  • Sugar alcohols
  • Glycerol
  • Aldoses
  • D-Glucose
• Carboxylic acids
  • Acetate

I could lump glycerol and glucose into one category or keep them in two separate ones.

[hgscott]

Here are the line graphs for the different metrics:

I'm a little unsure if the way I was calculating the degree of reduction.

Here's the code I'm using now:

c_num = met.elements["C"]
h = met.elements["H"] / c_num if 'H' in met.elements else 0
o = met.elements["O"] / c_num if 'O' in met.elements else 0
dor = 4 + h - 2 * o

But it doesn't account for nitrogen:

And some of the results don't match up to what Westerhoff said- like for Acetate I got 3.5 instead of 4.

[hgscott]

How can I figure out where the different compounds enter central carbon metabolism?

There are few enough that I could do it in a very manual/knowledge-rich way.

[hgscott]

Daniel said that the way I am calculating degree of reduction now is fine (the difference between my value for acetate and Westerhoff's was that my acetate was CH3COO- instead of CH3COOH.

And he said that grouping by entry point is not necessary, but would be nice. Instead just group by classes, or could calculate distance between reaction set (not fluxes like I did in PCA, but the total list of reactions that are above a certain threshold).