Compression function of NetworkReducer

38 views
Skip to first unread message

Marina Kurtz

unread,
Feb 10, 2022, 8:06:34 AM2/10/22
to CellNetAnalyzer User Forum
Hi,

I use the NetworkReducer including the compression function. According to which criteria are reaction paths summarized there? I sometimes get compressed responses that are not connected to each other in the original model and are even in different subsystems. Is that correct/intended? Can I just separate combined reactions?

Best regards
Marina

Steffen

unread,
Feb 10, 2022, 8:36:42 AM2/10/22
to CellNetAnalyzer User Forum
Hi Marina,

reactions may sometimes be lumped even if there is no connecting metabolite.

For example, assume you have the following pathway:

A->B->C->D->E

and that there is another parallel reaction (or pathway) converting C to D.
The compression function would lump the first, the second and the fourth
reaction of the above pathway yielding a reaction with stoichiometry
A+D->E+C (although the original fourth reaction is not linked to metabolites
of the first or second reaction).

Sometimes this may be undesired, you can then enforce in the compression
routine (via its parameters) that reaction 4 is not lumped. In this case, only
the first and the second reaction would be lumped to A->C.

Hope this helps,
Steffen.

Marina Kurtz

unread,
Feb 14, 2022, 5:31:03 AM2/14/22
to CellNetAnalyzer User Forum
Hi Steffen,

Thank you for your quick response. I've tried a few compression parameter settings (rational, rmChoke, and rmCR) but none of them give me the result I want.
Do you know what parameter allows these reactions (from your example reaction 4) to not lump?

Best regards
Marina

Steffen

unread,
Feb 14, 2022, 6:01:54 AM2/14/22
to CellNetAnalyzer User Forum
Hi Marina,

please use the protect_reac  parameter of that function.

In the example you could set protect_reac=4 to avoid lumping
of reaction 4. Of course, normally you will have to run the compression
once before you can decide which reaction should not be (or you do not
want to be) lumped.

Best,
Steffen.

Marina Kurtz

unread,
Feb 14, 2022, 7:32:00 AM2/14/22
to CellNetAnalyzer User Forum
Hi Steffen,

thanks. This makes sense. I will try it out.
I have another question: Is there a random seed or something like this in the reduction code? I run the reduction several times without compression (each time with the same parameters) and I got different results.

Best regards
Marina

Reply all
Reply to author
Forward
0 new messages