Section: New Results

Axis 2: Metabolism and post-transcriptional regulation

Multi-objective metabolic mixed integer optimisation: with an application to yeast strain engineering

In a paper submitted and already available in bioRxiv (https://www.biorxiv.org/content/early/2018/11/22/476689), we explored the concept of multi-objective optimisation in the field of metabolic engineering when both continuous and integer decision variables are involved in the model. In particular, we proposed a multi-objective model which may be used to suggest reaction deletions that maximise and/or minimise several functions simultaneously. The applications may include, among others, the concurrent maximisation of a bioproduct and of biomass, or maximisation of a bioproduct while minimising the formation of a given by-product, two common requirements in microbial metabolic engineering. Production of ethanol by the widely used cell factory Saccharomyces cerevisiae was adopted as a case study to demonstrate the usefulness of the proposed approach in identifying genetic manipulations that improve productivity and yield of this economically highly relevant bioproduct. We did an in vivo validation and we could show that some of the predicted deletions exhibit increased ethanol levels in comparison with the wild-type strain. The multi-objective programming framework we developed, called Momo , is open-source and uses PolySCIP as underlying multi-objective solver. This is part of the work of Ricardo de Andrade, postdoc at Unversity of São Paulo with Roberto Marcondes, and in ERABLE. It is joint work with Susana Vinga, external collaborator of ERABLE and partner of the Inria Associated Team Compasso.

Metabolic shifts

With the increasing availability of so-called 'omics data – transcriptomics, proteomics, and metabolics – there has been growing interest in various ways of integrating them with the metabolic network. When the network is represented by a graph, 'omics data can guide the extraction of subnetworks of interest to find metabolic pathways or sets of related genes. Within the framework of constraint-based modelling, 'omics data can be used to improve the prediction of metabolic behaviour and to build context-specific metabolic models. One interesting application of metabolic reconstructions in conjunction with 'omics data is to use the two to understand metabolic shifts. When an organism encounters a change in environmental conditions, often a re-organisation of metabolism follows. Comparative measurements of gene expression and metabolite concentrations can be used to gain insight into these changes but this data is "structureless", meaning it lacks the information about how the metabolic components relate to each other. A metabolic network on the other hand contains this information, and can thus greatly benefit such an analysis. We developed a new method, called Moomin , that combines the results of a differential expression analysis comparing the gene expression levels in two different conditions with a metabolic network to produce a hypothesis of a metabolic shift. The idea is to use the network structure to define feasible global changes in metabolism. These changes are then scored based on the gene expression data with the goal of finding the change that best agrees with the observations. Finding the best-scoring change is formulated into an optimisation problem that can be solved using Mixed-Integer Linear Programming. This is part of the work of Henri Taneli Pusa, co-supervised by 3 members of ERABLE, whose manuscript was submitted to the reviewers and who should be defending his PhD in early February 2019. The paper on Moomin will be submitted soon, and the software then made available. Participated also in this work Mariana G. Ferrarini, postdoc at Insa and in ERABLE, and Ricardo Andrade, postdoc at University of São Paulo with Roberto Marcondes and in ERABLE.

Metabolic games

The PhD of Taneli also investigated game theory in the context of metabolism. Game theory is a branch of applied mathematics that deals with interacting rational agents with conflicting goals. When rationality is replaced with natural selection, evolutionary game theory can be used to explain the "decisions" taken by even microscopic organisms. The PhD manuscript presents the idea of a metabolic game, a game theoretical model for the prediction of metabolic behaviour. In contrast to Flux Balance Analysis, where the metabolic state is predicted using simple optimisation, a metabolic game takes into account the fact that optimality is influenced by the surrounding members of a microbial community. By changing the availability of nutrients, or secreting beneficial or harmful molecules, microbes essentially create their own environment and make optimal behaviour context-specific. A paper is submitted that reviews the literature that has applied game theory to the study of microbes, with a focus on metabolism and especially games derived using metabolic networks and constraint-based modelling. In the PhD manuscript, Taneli further explains the idea behind a metabolic game and discusses different aspects of defining such games: the choice of players, actions, and payoffs.