Zoom Logo

MFBM - Shared screen with speaker view
Jae Kyoung Kim
02:40:25
You can leave the question in this chatting room...
Bryan Hernandez
02:42:05
1. Given any (bio)chemical reaction network, does there always exist as weakly reversible and deficiency zero network translation? 2. Also a translated network is not always mass action, complex balancing is not ensured. In case that there is no complex balance equilibrium, can we say something about the existence of the stationary distribution about the original stochastic BRN? Thanks a lot. :)
Editha Jose
02:47:53
Nice talk! But there are also cases that many translated networks of the original will be weakly reversible and deficiency zero? Is that right? Do you have to choose?
Hyukpyo Hong
02:56:20
@EdithaYou are right, there are multiple choice of transalted network with the desired properties.Since a translated network has to further satisfy the 'propensity factorization' condition, we have to choose appropriate one.
Editha Jose
03:00:23
I see, nice work! Thanks @Hyukpyo :)
Hyukpyo Hong
03:02:30
Thanks, please feel free to email me (hphong@kaist.ac.kr) when you have any question and comment :)
Brian Munsky (he/him)
03:15:05
@Zhou - great talk. Regarding time delays, it could be really nice to add a delay via additional reactions for the activation and decay of the GFP reporter and ask how sensitive is the estimator under different maturation and degradation times. I’d like to try your idea on some of the data I will show in my talk — let’s chat in the break!
Zhou Fang
03:22:24
@Brian, I see. It sounds really great! Let’s discuss it in the break!
SJC
03:26:06
@SamHow do you model kappa(x,y) s?Are those location-dependent reaction rates?If so, how can we obtain kappa(x,y) from a reaction rate of ODE?
Brian Munsky (he/him)
03:26:10
Could you repeat how the k_ij terms are defined and how this depends on the discretization? I presume it drops off as the i and j are separated by greater distances.
Jae Kyoung Kim
03:32:51
Brian you can begin sharing your scrreen
Jae Kyoung Kim
03:33:35
Brian/Zhou: FYI: Here is our recent work about inferring time delay distribution … https://academic.oup.com/bioinformatics/article/36/2/586/5538987?login=true It would be nice if you can merge the hybrid approximation here…
Samuel A Isaacson
03:37:28
@SJC and @Brian Munsky there are lots of possible choices for the bimolecular reaction functions. We usually use Doi models, where molecules react with fixed probabilities per time once they are sufficiently close (so k(x,y) is an indicator function, with a reaction radius at which it transitions from zero to a fixed rate). For some membrane interactions we’ve used Gaussian interactions that have been derived from more microscopic polymer models. In all cases these are short-range interactions, so go to zero over short length scales (comparable to molecular sizes).
Samuel A Isaacson
03:41:20
@SJC, there are different ways to determine parameters. We have recent work using SPR to directly fit microscopic particle model parameters, but it is also common to chose those parameters so that in the fast diffusion limit one recovers the rate constant of a well-mixed model.
SJC
03:47:36
@SamThank you for the reply. Really appreciate it!
Rodrigo A. García
04:00:54
Thank you for the great talks!