Radivojevic, T. | Search bioRxiv

bioRxiv2025

Prediction of metabolic dynamics through deep learning and high-throughput multiomics data

Synthetic biologys remarkable potential to tackle important societal problems is held back by our inability to predictably engineer biological systems. Here, we collected one of the largest public multiomics synthetic biology datasets generated to date, and used it to train a novel deep learning algorithm able to predict product and metabolic dynamics with great accuracy, starting to approach the predictive capabilities found in physics and chemistry. We were able to predict production time series with 90-99% accuracy, and final production with 96% accuracy. Further, we were able to produce good predictions for a majority of extracellular metabolites, and twenty different intracellular metabolites. These predictions were provided for a target of industrial relevance: a non-model yeast (Pichia kudriavzevii) engineered to produce large amounts of malonic acid, a desirable biomanufacturing target. This approach is generally applicable to any host, pathway, and product because all required knowledge is inferred from experimental data.

bioengineering↗

bioRxiv2023

BayFlux: A Bayesian method to quantify metabolic Fluxes and their uncertainty at the genome scale

Metabolic fluxes, the number of metabolites traversing each biochemical reaction in a cell per unit time, are crucial for assessing and understanding cell function. 13C Metabolic Flux Analysis (13C MFA) is considered to be the gold standard for measuring metabolic fluxes. 13C MFA typically works by leveraging extracellular exchange fluxes as well as data from 13C labeling experiments to calculate the flux profile which best fit the data for a small, central carbon, metabolic model. However, the nonlinear nature of the 13C MFA fitting procedure means that several flux profiles fit the experimental data within the experimental error, and traditional optimization methods offer only a partial or skewed picture, especially in "non-gaussian" situations where multiple very distinct flux regions fit the data equally well. Here, we present a method for flux space sampling through Bayesian inference (BayFlux), that identifies the full distribution of fluxes compatible with experimental data for a comprehensive genome-scale model. This Bayesian approach allows us to accurately quantify uncertainty in calculated fluxes. We also find that, surprisingly, the genome-scale model of metabolism produces narrower flux distributions (reduced uncertainty) than the small core metabolic models traditionally used in 13C MFA. The different results for some reactions when using genome-scale models vs core metabolic models advise caution in assuming strong inferences from 13C MFA since the results may depend significantly on the completeness of the model used. Based on BayFlux, we developed and evaluated novel methods (P-13C MOMA and ROOM) to predict the biological results of a gene knockout, that improve on the traditional MOMA and ROOM methods. We provide an open source Python implementation of BayFlux at https://github.com/JBEI/bayflux. Author summary13C MFA practitioners know that modeling results can be sensitive to minor modifications of the metabolic model. Certain parts of the metabolic model that are not well mapped to a molecular mechanism (e.g. drains to biomass or ATP maintenance) can have an inordinate impact on the final fluxes. The only way to ascertain the validity of the model is by checking that the result does not significantly differ from previously observed flux profiles. However, that approach diminishes the possibility of discovering truly novel flux profiles. Because of this strong dependence on metabolic model details, it would be very useful to have a systematic and repeatable way to produce these metabolic models. And indeed there is one: genome-scale metabolic models can be systematically obtained from genomic sequences, and represent all the known genomically encoded metabolic information. However, these models are much larger than the traditionally used central carbon metabolism models. Hence, the number of degrees of freedom of the model (fluxes) significantly exceeds the number of measurements (metabolite labeling profiles and exchange fluxes). As a result, one expects many flux profiles compatible with the experimental data. The best way to represent these is by identifying all fluxes compatible with the experimental data. Our novel method BayFlux, based on Bayesian inference and Markov Chain Monte Carlo sampling, provides this capability. Interestingly, this approach leads to the observation that traditional optimization approaches can significantly overestimate flux uncertainty, and that genome-scale models of metabolism produce narrower flux distributions than the small core metabolic models that are traditionally used in 13C MFA. Furthermore, we show that the extra information provided by this approach allows us to improve knockout predictions, compared to traditional methods. Although the method scales well with more reactions, improvements will be needed to tackle the large metabolic models found in microbiomes and human metabolism.

systems biology↗

bioRxiv2020

Multiomics data collection, visualization, and utilization for guiding metabolic engineering

Biology has changed radically in the past two decades, growing from a purely descriptive science into also a design science. The availability of tools that enable the precise modification of cells, as well as the ability to collect large amounts of multimodal data, open the possibility of sophisticated bioengineering to produce fuels, specialty and commodity chemicals, materials, and other renewable bioproducts. However, despite new tools and exponentially increasing data volumes, synthetic biology cannot yet fulfill its true potential due to our inability to predict the behavior of biological systems. Here, we present a set of tools that, combined, provide the ability to store, visualize and leverage these data to predict the outcome of bioengineering efforts.

synthetic biology↗

bioRxiv2019

Predictive engineering and optimization of tryptophan metabolism in yeast through a combination of mechanistic and machine learning models

In combination with advanced mechanistic modeling and the generation of high-quality multi-dimensional data sets, machine learning is becoming an integral part of understanding and engineering living systems. Here we show that mechanistic and machine learning models can complement each other and be used in a combined approach to enable accurate genotype-to-phenotype predictions. We use a genome-scale model to pinpoint engineering targets and produce a large combinatorial library of metabolic pathway designs with different promoters which, once phenotyped, provide the basis for machine learning algorithms to be trained and used for new design recommendations. The approach enables successful forward engineering of aromatic amino acid metabolism in yeast, with the new recommended designs improving tryptophan production by up to 17% compared to the best designs used for algorithm training, and ultimately producing a total increase of 106% in tryptophan accumulation compared to optimized reference designs. Based on a single high-throughput data-generation iteration, this study highlights the power of combining mechanistic and machine learning models to enhance their predictive power and effectively direct metabolic engineering efforts.

bioengineering↗