Artificial Metabolic Networks: enabling neural computation with metabolic networks

Metabolic networks have largely been exploited as mechanistic tools to predict the behavior of microorganisms with a defined genotype in different environments. However, flux predictions by constraint-based modeling approaches are limited in quality unless labor intensive experiments including the measurement of media intake fluxes, are performed. Using machine learning instead of an optimization of biomass flux – on which most existing constraint-based methods are based – provides ways to improve flux and growth rate predictions. In this paper, we show how Recurrent Neural Networks can surrogate constraint-based modeling and make metabolic networks suitable for backpropagation and consequently be used as an architecture for machine learning. We refer to our hybrid mechanistic and neural network – models as Artificial Metabolic Networks (AMN). We showcase AMN and illustrate its performance with an experimental dataset of Escherichia coli growth rates in 73 different media compositions. We reach a regression coefficient of R=0.78 on crossvalidation sets. We expect AMNs to provide easier discovery of metabolic insights and prompt new biotechnological applications.

[1]  Ranjan Anantharaman,et al.  Stably Accelerating Stiff Quantitative Systems Pharmacology Models: Continuous-Time Echo State Networks as Implicit Machine Learning , 2021, bioRxiv.

[2]  Bryan D. Bryson,et al.  Artificial neural networks enable genome-scale simulations of intracellular signaling , 2021, Nature Communications.

[3]  J. Szymański,et al.  Advances in flux balance analysis by integrating machine learning and mechanism-based models , 2021, Computational and structural biotechnology journal.

[4]  Qin Ma,et al.  A graph neural network model to estimate cell-wise metabolic flux using single-cell RNA-seq data , 2021, Genome research.

[5]  J. Faulon,et al.  In silico, in vitro, and in vivo machine learning in synthetic biology and metabolic engineering. , 2021, Current opinion in chemical biology.

[6]  Joshua E. Lewis,et al.  Integration of machine learning and genome-scale metabolic modeling identifies multi-omics biomarkers for radiation resistance , 2020, Nature Communications.

[7]  Claudio Angione,et al.  A mechanism-aware and multiomic machine-learning pipeline characterizes yeast cell growth , 2020, Proceedings of the National Academy of Sciences.

[8]  John T. Nardini,et al.  Biologically-informed neural networks guide mechanistic modeling from sparse experimental data , 2020, PLoS Comput. Biol..

[9]  U. Feige,et al.  The Simplex Algorithm , 2020, Linear Algebra and Optimization with Applications to Machine Learning.

[10]  Pablo Carbonell,et al.  Opportunities at the Intersection of Synthetic Biology, Machine Learning, and Automation. , 2019, ACS synthetic biology.

[11]  Claudio Angione,et al.  Machine and deep learning meet genome-scale metabolic modeling , 2019, PLoS Comput. Biol..

[12]  Shuai Li,et al.  A survey on projection neural networks and their applications , 2019, Appl. Soft Comput..

[13]  Habib N. Najm,et al.  Workshop Report on Basic Research Needs for Scientific Machine Learning: Core Technologies for Artificial Intelligence , 2018 .

[14]  Toshiyuki Yamane,et al.  Recent Advances in Physical Reservoir Computing: A Review , 2018, Neural Networks.

[15]  Diogo M. Camacho,et al.  Next-Generation Machine Learning for Biological Networks , 2018, Cell.

[16]  Ilias Tagkopoulos,et al.  Multi-omics integration accurately predicts cellular state in unexplored conditions for Escherichia coli , 2016, Nature Communications.

[17]  Tianqi Chen,et al.  XGBoost: A Scalable Tree Boosting System , 2016, KDD.

[18]  Yinjie J. Tang,et al.  Facilitate Collaborations among Synthetic Biology, Metabolic Engineering and Machine Learning , 2016 .

[19]  Georg Regensburger,et al.  Resource allocation in metabolic networks: kinetic optimization and approximations by FBA. , 2015, Biochemical Society transactions.

[20]  W. Wiechert,et al.  How to measure metabolic fluxes: a taxonomic guide for (13)C fluxomics. , 2015, Current opinion in biotechnology.

[21]  Jinde Cao,et al.  A new neural network for solving quadratic programming problems with equality and inequality constraints , 2014, Math. Comput. Simul..

[22]  Joshua A. Lerman,et al.  COBRApy: COnstraints-Based Reconstruction and Analysis for Python , 2013, BMC Systems Biology.

[23]  D. Fell,et al.  Is maximization of molar yield in metabolic networks favoured by evolution? , 2008, Journal of theoretical biology.

[24]  M. A. de Menezes,et al.  Intracellular crowding defines the mode and sequence of substrate uptake by Escherichia coli and constrains its metabolic activity , 2007, Proceedings of the National Academy of Sciences.

[25]  Nezam Mahdavi-Amiri,et al.  An efficient simplified neural network for solving linear and quadratic programming problems , 2006, Appl. Math. Comput..

[26]  B. Palsson,et al.  Thirteen Years of Building Constraint-Based In Silico Models of Escherichia coli , 2003, Journal of bacteriology.

[27]  B. Palsson,et al.  Metabolic capabilities of Escherichia coli: I. synthesis of biosynthetic precursors and cofactors. , 1993, Journal of theoretical biology.

[28]  Jun Wang,et al.  Recurrent neural networks for linear programming: Analysis and design principles , 1992, Comput. Oper. Res..

[29]  Jorge Nocedal,et al.  On the limited memory BFGS method for large scale optimization , 1989, Math. Program..

[30]  J. J. Hopfield,et al.  “Neural” computation of decisions in optimization problems , 1985, Biological Cybernetics.

[31]  Diana M. Hendrickx,et al.  MetDFBA: incorporating time-resolved metabolomics measurements into dynamic flux balance analysis. , 2015, Molecular bioSystems.

[32]  Skipper Seabold,et al.  Statsmodels: Econometric and Statistical Modeling with Python , 2010, SciPy.