eduzhai > Applied Sciences > Engineering >

Prediction of fitness in bacteria with causal jump dynamic mode decomposition

  • Save

... pages left unread,continue reading

Document pages: 8 pages

Abstract: In this paper, we consider the problem of learning a predictive model forpopulation cell growth dynamics as a function of the media conditions. We firstintroduce a generic data-driven framework for training operator-theoreticmodels to predict cell growth rate. We then introduce the experimental designand data generated in this study, namely growth curves of Pseudomonas putida asa function of casein and glucose concentrations. We use a data driven approachfor model identification, specifically the nonlinear autoregressive (NAR) modelto represent the dynamics. We show theoretically that Hankel DMD can be used toobtain a solution of the NAR model. We show that it identifies a constrainedNAR model and to obtain a more general solution, we define a causal state spacesystem using 1-step,2-step,...,{ tau}-step predictors of the NAR model andidentify a Koopman operator for this model using extended dynamic modedecomposition. The hybrid scheme we call causal-jump dynamic modedecomposition, which we illustrate on a growth profile or fitness predictionchallenge as a function of different input growth conditions. We show that ourmodel is able to recapitulate training growth curve data with 96.6 accuracyand predict test growth curve data with 91 accuracy.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×