OptKnock and non-growth associated knockouts

[Chris]

Hello all,

I've run OptKnock a number of times without finding a growth-
associated strain design, but with every set of conditions, optknock
finds an optimal solution with a high objective value that doesn't
actually coincide with optimal growth. So, for example, optknock says
my target is produced at 20, which is the theoretical optimum for my
conditions, but the growth rate for this solution is right at my
minimum growth constraint. If I take the solution and test it, for
example using analyzeOptKnock(), I find that the deletion is not
growth associating, and my production is ~0 at the true optimal
growth.

Is this typical behavior for the optknock algorithm when no growth-
associating deletion is possible? If that is the case, I can work with
that, but I am worried by the fact that it seems to be settling at a
non-optimal growth rate for the solution strain. I have checked and re-
checked my constraints, and AFAIK they are all properly handled in
constrOpt (for optknock) or as part of the model (for analyzeOptKnock
or other testing).

Maybe I'm worrying about nothing? Thanks for any help you can offer.

Chris

[Ronan Fleming]

Hi Chris,
at a glance, this might be because the lower line on your production
envelope is vertical. You need both to be sloping in the same
direction. Adam Feist at UCSD has much more experience with this than
I though.
Ronan

--
--
Mr. Ronan MT Fleming B.V.M.S. Dip. Math. Ph.D.
-----------------------------------------------------------------
Independent Group Leader,
Center for Systems Biology,
University of Iceland,
Sturlugata 8,
101 Reykjavik,
Iceland.
http://www.hi.is/~rfleming
http://systemsbiology.hi.is/
Ph:  +354 618 6245
-----------------------------------------------------------------

[Josh]

Hi,

I just started using OptKnock this week and I am having a similar
issue to what Chris mentioned back in July. I have been attempting to
find a 3 deletion solution for overproducing a certain metabolite in
the genome-scale E. coli iJR904 model. As he mentioned, the solution
returned by the optknock function is not growth-associated and the
true production of the target metabolite is 0 at optimal biomass.

Did anyone ever figure out if this is typical behavior when their is
no growth-associated solution? Or is it possible to get this kind of
result when there is another growth associated solution?

I am concerned about the usefulness of this function if it can predict
a non-growth associated solution when other growth-associated
solutions are available. However, if a non-growth associated solution
is only returned when growth associated solutions are not possible, it
would be easy for me to change the number of deletions or some other
constraints.

Thanks,
Josh

On Aug 8, 5:02 am, Ronan Fleming <ronan.mt.flem...@gmail.com> wrote:
> Hi Chris,
> at a glance, this might be because the lower line on your production
> envelope is vertical. You need both to be sloping in the same
> direction. Adam Feist at UCSD has much more experience with this than
> I though.
> Ronan
>

> Iceland.http://www.hi.is/~rfleminghttp://systemsbiology.hi.is/

[Adam Feist]

Josh,

Have a look at these two publications, I think they may give you some
insight into your issue:

Feist AM, Zielinski DC, Orth JD, Schellenberger J, Herrgard MJ,
Palsson BO. 2010. Model-driven evaluation of the production
potential for growth-coupled products of Escherichia coli.. Metabolic
engineering. 12(3):173-86

Predicting Metabolic Engineering Knockout Strategies for Chemical
Production: Accounting for Competing Pathways.
N. Tepper, T. Shlomi
Journal: Bioinformatics, 2009, doi:10.1093

The issue may be to "non-unique" solutions.

Best,
Adam

[Josh]

Those papers are very helpful. Thank you!