Succinate flux in Optnock Tutorial

[alexander vindel]

Hello,

I am quite new to metabolic engineering and I have been working in trying to develop an algorithm to solve OptKnock using python and gurbobi. I am basing on the Optnock tutorial found in COBRA. In the tutorial the best Knock out set is conformed by 'PFL' and 'RPI" giving a succinate overproduction of 2.75, however when I run FBA without those two reactions the value I get is 2.3. 

I run FBA using the same biological assumptions as the tutorial with the FBA constraints adding a binary variable in the LB and UB:

Max v_biomass

S_ij * v_j = 0:

LB_j * y_j <= v_j :

v_j <= UB_j * y_j

Is there something I am not considering in the FBA (inner model) that limits the values I get from the value reported in the tutorial? I'd appreciate the help.

[caroline Fukuda]

Hi alexander, 

when you're running the FBA, are you assuming the succinate exchange as the objective function?

--

---
You received this message because you are subscribed to the Google Groups "COBRA Toolbox" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cobra-toolbo...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cobra-toolbox/7c48a02d-55b7-4fe0-a40d-5a64ec97c198n%40googlegroups.com.

[alexander vindel]

Hello Carolin,

I'm assuming the biomass as the objective function, much like it would be for the inner problem where the objective function is the cellular growth.

[alexander vindel]

I just ran an instance where the succinate exchange is the objective function, the value I get is 15.76, and the cellular growth (biomass) = 0.0 

[caroline Fukuda]

I don't know how to explain it right because I don't have much experience.

The FBA analysis gives you the maximization for the objective function, because for the other reactions there are a set of different solutions

To view this discussion on the web visit https://groups.google.com/d/msgid/cobra-toolbox/f6a66bd4-5bc2-403d-86c1-cb2e5a69f7afn%40googlegroups.com.

[alexander vindel]

I'll keep tinkering around to try to understand what's going on, thanks!!

[Hoang Dinh]

Hi, the objective used in OptKnock is roughly max production (actually, it is max production coupled with max biomass via bilevel linear programming). Thus, to get the value 2.75, you need to run two FBA problems back-to-back:

(1) Max biomass

(2) Max v(product) subject to v(biomass) = vmax(biomass)

When you get an OptKnock solution, you would want to check if your solution is robust by running the following two FBA problems back-to-back:

(1) Max biomass

(2) Min v(product) subject to v(biomass) = vmax(biomass)

When you minimizing biomass, you will discover that there is a competing pathway that can lead to the same growth-coupled max biomass production.

[alexander vindel]

Thank you so much for your answer. 

Just to clarify, first I'd run:

max v(product)

st.

1) max v(biomass)

    1.1) S.v = 0

    1.2) LB(yi) <= vi <= UB(yi)

2) sum(y_i) <= knockouts

To get the knock out solutions and then I'd run:

min v(product)

st.

1)v(biomass) = vmax(biomass) (vmax_biomass would be the value you get from the wildtpe?)

[Hoang Dinh]

To get the KO solutions, it should be coming from OptKnock.

Once you get the KOs, you should run

(Problem 1)

max vBiomass

s.t. S.v = 0

v(RxnKO) = 0

=> You obtain vBiomassMaxProblem1

(Problem 2)

max/min vProduct (max will get you OptKnock value, min will be useful to check for strain robustness. You can also search for the concept of production envelope regarding this)

s.t. S.v = 0

v(RxnKO) = 0

vBiomass = vBiomassMaxProblem1

vmaxBiomass is for the mutant.