Integration of metabolic models in biorefinery designs using superstructure optimisation
Abstract
In this work the integration of metabolic (structured) models, which describe the entire metabolic network of microorganisms, into a superstructure optimisation framework is presented. These models are of particular interest because they can predict the outcome of a fermentation with different substrates, and without the need of experimental data. In this contribution, a workflow for 2 types of structured models is described: i) genome-scale metabolic models (GEMs) and ii) community models where product yields can be influenced by environmental factors (e.g., pH). To showcase this methodology a simple case study is presented of a superstructure optimisation problem. With this case study it is demonstrated that the described workflow can aid the design of novel biorefineries, by screening potential cultures using metabolic models without the need of experimental data.
Full text
Advancing biorefinery design through the integration of metabolic models in superstructure optimization Lucas Van der Hauwaert, Alberte Regueira, Miguel Mauricio-Iglesias CRETUS Institute. Department of Chemical Engineering, Universidade de Santiago de Compostela. mail: lucas.vanderhauw[email protected] #escape2023
The biorefinery concept 2
Challenges of biorefinery design •Feedstock Availability and Variability •Techno-economic viability & Scale up Product Diversification Linking substrates to potential products 3 How do we assure investors to develop biorefineries?
4 Galan et al. (2021) Superstructure optimisation: Sorbitol and Xylitol Objective
5 Galan et al. (2021) Objective Superstructure optimisation: succinate acid
6 Galan et al. (2021) 4 distinct microorganisms with well established fermentation outcomes BUT Enormous Diversity of microorganisms for bioreactors and the potential substrates they can consume Representation bio-reactor stage Bio chemical conversion of substrates to products
? Getting the bioreactions in superstructures Research question: How to integrate bioreactions of microorganisms in the most effective and efficient way in the superstructure? Or How to unlock the full potential of microorganism in a superstructure framework? Goal: Can we explore completely new designs linking new microorganisms to substrates and products? 7
Advantages •Predictive from omics data •Able to handle multiple substrates Representing Bioreactors in Superstructures Metabolic models 8
Considered biotransformations in bioreactors Pure cultures Genome Metabolic Models (GEMS) Open mixed cultures Bioenergetic model Recreation of the metabolic network, starting from the genome pH Recreation of metabolic networks, But dependent on environmental influences 9
Case study P. freudenreichii P. avidum P. acidipropionici Mixed culture pH P. acnes P. propionicum Input Interval Selective separation Interval Reactor Interval Propionate Acetate Output interval Destillation 3 Concentration Interval Liq –Liq + Distillation Destillation 2 waste D-Glucose D-Fructose Sucrose Maltose Dextrin L-Lactate Glycerol Xylose 28 different combinations of substrate –microorganism Objective: EBIT = - OPEX GREV 16 How to identify substrates: 1. Identify Exchange reactions in the GEM 2. Filter out proteins 3. A minimum production of the target products required 4. Final manual selection
Results P. acnes Input Interval Selective separation Interval Reactor Interval Propionate Acetate Output interval Destillation 3 Concentration Interval Liq –Liq + Distillation waste L-Lactate Revenue: 27 418 € EBIT: 3466 € Substrate load: 24 000 kg Cost raw material : 11040 € Propionate: 12 221 kg Acetate: 453 kg OPEX : 786 € Propionate: 12 099 kg Acetate: 444 kg OPEX : 6849 € Propionate: 11 978 kg Acetate: 440 kg OPEX : 1605 € 17
Conclusion Case studies demonstrate how superstructure optimization using metabolic models can identify economically viable combinations of substrates and microorganisms. An efficient workflow was created to integrate metabolic models for representing bioreactor processes in superstructure optimization problems The surrogate models are capable of 1) Predicting operation conditions for optimal yields 2) Predictive from genome data 3) Predicting yields from various substrates 18
THANK YOU HAVE A NICE DAY EMAIL [email protected] Acknowledgments This work was supported by project ALQUIMIA (PID2019-110993RJ-I00) funded by the Agencia Estatal de Investigación Alquimia: Proyecto de I- D-i Programa Retos de la sociedad modalidad Jovenes investigadores convocatoria. 19
EXTRA 20
Representation of unit processes Flow divider Mixer Mixer utilities (chemicals) (Bio)Reactor Separator Flow of mass Flow energy Next process unit or waste OPEX Quaglia et al (2015), Bertran et al. (2017) €€€ € 21