Sequence-based Network Completion Reveals the Integrality of Missing Reactions in Metabolic Networks

Genome-scale metabolic models are central in connecting genotypes to metabolic phenotypes. However, even for well studied organisms, such as Escherichia coli, draft networks do not contain a complete biochemical network. Missing reactions are referred to as gaps. These gaps need to be filled to enable functional analysis, and gap-filling choices influence model predictions. To investigate whether functional networks existed where all gap-filling reactions were supported by sequence similarity to annotated enzymes, four draft networks were supplemented with all reactions from the Model SEED database for which minimal sequence similarity was found in their genomes. Quadratic programming revealed that the number of reactions that could partake in a gap-filling solution was vast: 3,270 in the case of E. coli, where 72% of the metabolites in the draft network could connect a gap-filling solution. Nonetheless, no network could be completed without the inclusion of orphaned enzymes, suggesting that parts of the biochemistry integral to biomass precursor formation are uncharacterized. However, many gap-filling reactions were well determined, and the resulting networks showed improved prediction of gene essentiality compared with networks generated through canonical gap filling. In addition, gene essentiality predictions that were sensitive to poorly determined gap-filling reactions were of poor quality, suggesting that damage to the network structure resulting from the inclusion of erroneous gap-filling reactions may be predictable.     

The RAVEN Toolbox and Its Use for Generating a Genome-scale Metabolic Model for Penicillium chrysogenum

We present the RAVEN (Reconstruction, Analysis and Visualization of Metabolic Networks) Toolbox: a software suite that allows for semi-automated reconstruction of genome-scale models. It makes use of published models and/or the KEGG database, coupled with extensive gap-filling and quality control features. The software suite also contains methods for visualizing simulation results and omics data, as well as a range of methods for performing simulations and analyzing the results. The software is a useful tool for system-wide data analysis in a metabolic context and for streamlined reconstruction of metabolic networks based on protein homology. The RAVEN Toolbox workflow was applied in order to reconstruct a genome-scale metabolic model for the important microbial cell factory Penicillium chrysogenum Wisconsin54-1255. The model was validated in a bibliomic study of in total 440 references, and it comprises 1471 unique biochemical reactions and 1006 ORFs. It was then used to study the roles of ATP and NADPH in the biosynthesis of penicillin, and to identify potential metabolic engineering targets for maximization of penicillin production.     

PSAMM: A Portable System for the Analysis of Metabolic Models

The genome-scale models of metabolic networks have been broadly applied in phenotype prediction, evolutionary reconstruction, community functional analysis, and metabolic engineering. Despite the development of tools that support individual steps along the modeling procedure, it is still difficult to associate mathematical simulation results with the annotation and biological interpretation of metabolic models. In order to solve this problem, here we developed a Portable System for the Analysis of Metabolic Models (PSAMM), a new open-source software package that supports the integration of heterogeneous metadata in model annotations and provides a user-friendly interface for the analysis of metabolic models. PSAMM is independent of paid software environments like MATLAB, and all its dependencies are freely available for academic users. Compared to existing tools, PSAMM significantly reduced the running time of constraint-based analysis and enabled flexible settings of simulation parameters using simple one-line commands. The integration of heterogeneous, model-specific annotation information in PSAMM is achieved with a novel format of YAML-based model representation, which has several advantages, such as providing a modular organization of model components and simulation settings, enabling model version tracking, and permitting the integration of multiple simulation problems. PSAMM also includes a number of quality checking procedures to examine stoichiometric balance and to identify blocked reactions. Applying PSAMM to 57 models collected from current literature, we demonstrated how the software can be used for managing and simulating metabolic models. We identified a number of common inconsistencies in existing models and constructed an updated model repository to document the resolution of these inconsistencies.     

Finding elementary flux modes in metabolic networks based on flux balance analysis and flux coupling analysis: application to the analysis of Escherichia coli metabolism

Elementary modes (EMs) are steady-state metabolic flux vectors with minimal set of active reactions. Each EM corresponds to a metabolic pathway. Therefore, studying EMs is helpful for analyzing the production of biotechnologically important metabolites. However, memory requirements for computing EMs may hamper their applicability as, in most genome-scale metabolic models, no EM can be computed due to running out of memory. In this study, we present a method for computing randomly sampled EMs. In this approach, a network reduction algorithm is used for EM computation, which is based on flux balance-based methods. We show that this approach can be used to recover the EMs in the medium- and genome-scale metabolic network models, while the EMs are sampled in an unbiased way. The applicability of such results is shown by computing “estimated” control-effective flux values in Escherichia coli metabolic network.     

GrowMatch: An Automated Method for Reconciling In Silico/In Vivo Growth Predictions

Genome-scale metabolic reconstructions are typically validated by comparing in silico growth predictions across different mutants utilizing different carbon sources with in vivo growth data. This comparison results in two types of model-prediction inconsistencies; either the model predicts growth when no growth is observed in the experiment (GNG inconsistencies) or the model predicts no growth when the experiment reveals growth (NGG inconsistencies). Here we propose an optimization-based framework, GrowMatch, to automatically reconcile GNG predictions (by suppressing functionalities in the model) and NGG predictions (by adding functionalities to the model). We use GrowMatch to resolve inconsistencies between the predictions of the latest in silico Escherichia coli (iAF1260) model and the in vivo data available in the Keio collection and improved the consistency of in silico with in vivo predictions from 90.6% to 96.7%. Specifically, we were able to suggest consistency-restoring hypotheses for 56/72 GNG mutants and 13/38 NGG mutants. GrowMatch resolved 18 GNG inconsistencies by suggesting suppressions in the mutant metabolic networks. Fifteen inconsistencies were resolved by suppressing isozymes in the metabolic network, and the remaining 23 GNG mutants corresponding to blocked genes were resolved by suitably modifying the biomass equation of iAF1260. GrowMatch suggested consistency-restoring hypotheses for five NGG mutants by adding functionalities to the model whereas the remaining eight inconsistencies were resolved by pinpointing possible alternate genes that carry out the function of the deleted gene. For many cases, GrowMatch identified fairly nonintuitive model modification hypotheses that would have been difficult to pinpoint through inspection alone. In addition, GrowMatch can be used during the construction phase of new, as opposed to existing, genome-scale metabolic models, leading to more expedient and accurate reconstructions.     

Physiologically Shrinking the Solution Space of a Saccharomyces cerevisiae Genome-Scale Model Suggests the Role of the Metabolic Network in Shaping Gene Expression Noise

Sampling the solution space of genome-scale models is generally conducted to determine the feasible region for metabolic flux distribution. Because the region for actual metabolic states resides only in a small fraction of the entire space, it is necessary to shrink the solution space to improve the predictive power of a model. A common strategy is to constrain models by integrating extra datasets such as high-throughput datasets and C¹³-labeled flux datasets. However, studies refining these approaches by performing a meta-analysis of massive experimental metabolic flux measurements, which are closely linked to cellular phenotypes, are limited. In the present study, experimentally identified metabolic flux data from 96 published reports were systematically reviewed. Several strong associations among metabolic flux phenotypes were observed. These phenotype-phenotype associations at the flux level were quantified and integrated into a Saccharomyces cerevisiae genome-scale model as extra physiological constraints. By sampling the shrunken solution space of the model, the metabolic flux fluctuation level, which is an intrinsic trait of metabolic reactions determined by the network, was estimated and utilized to explore its relationship to gene expression noise. Although no correlation was observed in all enzyme-coding genes, a relationship between metabolic flux fluctuation and expression noise of genes associated with enzyme-dosage sensitive reactions was detected, suggesting that the metabolic network plays a role in shaping gene expression noise. Such correlation was mainly attributed to the genes corresponding to non-essential reactions, rather than essential ones. This was at least partially, due to regulations underlying the flux phenotype-phenotype associations. Altogether, this study proposes a new approach in shrinking the solution space of a genome-scale model, of which sampling provides new insights into gene expression noise.     

Metabolic network modeling and simulation for drug targeting and discovery

Systems biology has greatly contributed toward the analysis and understanding of biological systems under various genotypic and environmental conditions on a much larger scale than ever before. One of the applications of systems biology can be seen in unraveling and understanding complicated human diseases where the primary causes for a disease are often not clear. The in silico genome-scale metabolic network models can be employed for the analysis of diseases and for the discovery of novel drug targets suitable for treating the disease. Also, new antimicrobial targets can be discovered by analyzing, at the systems level, the genome-scale metabolic network of pathogenic microorganisms. Such applications are possible as these genome-scale metabolic network models contain extensive stoichiometric relationships among the metabolites constituting the organism's metabolism and information on the associated biophysical constraints. In this review, we highlight applications of genome-scale metabolic network modeling and simulations in predicting drug targets and designing potential strategies in combating pathogenic infection. Also, the use of metabolic network models in the systematic analysis of several human diseases is examined. Other computational and experimental approaches are discussed to complement the use of metabolic network models in the analysis of biological systems and to facilitate the drug discovery pipeline.     

Reconciliation of genome-scale metabolic reconstructions for comparative systems analysis

In the past decade, over 50 genome-scale metabolic reconstructions have been built for a variety of single- and multi- cellular organisms. These reconstructions have enabled a host of computational methods to be leveraged for systems-analysis of metabolism, leading to greater understanding of observed phenotypes. These methods have been sparsely applied to comparisons between multiple organisms, however, due mainly to the existence of differences between reconstructions that are inherited from the respective reconstruction processes of the organisms to be compared. To circumvent this obstacle, we developed a novel process, termed metabolic network reconciliation, whereby non-biological differences are removed from genome-scale reconstructions while keeping the reconstructions as true as possible to the underlying biological data on which they are based. This process was applied to two organisms of great importance to disease and biotechnological applications, Pseudomonas aeruginosa and Pseudomonas putida, respectively. The result is a pair of revised genome-scale reconstructions for these organisms that can be analyzed at a systems level with confidence that differences are indicative of true biological differences (to the degree that is currently known), rather than artifacts of the reconstruction process. The reconstructions were re-validated with various experimental data after reconciliation. With the reconciled and validated reconstructions, we performed a genome-wide comparison of metabolic flexibility between P. aeruginosa and P. putida that generated significant new insight into the underlying biology of these important organisms. Through this work, we provide a novel methodology for reconciling models, present new genome-scale reconstructions of P. aeruginosa and P. putida that can be directly compared at a network level, and perform a network-wide comparison of the two species. These reconstructions provide fresh insights into the metabolic similarities and differences between these important Pseudomonads, and pave the way towards full comparative analysis of genome-scale metabolic reconstructions of multiple species.     

Genome-scale model for Clostridium acetobutylicum: Part I. Metabolic network resolution and analysis

A genome-scale metabolic network reconstruction for Clostridium acetobutylicum (ATCC 824) was carried out using a new semi-automated reverse engineering algorithm. The network consists of 422 intracellular metabolites involved in 552 reactions and includes 80 membrane transport reactions. The metabolic network illustrates the reliance of clostridia on the urea cycle, intracellular L-glutamate solute pools, and the acetylornithine transaminase for amino acid biosynthesis from the 2-oxoglutarate precursor. The semi-automated reverse engineering algorithm identified discrepancies in reaction network databases that are major obstacles for fully automated network-building algorithms. The proposed semi-automated approach allowed for the conservation of unique clostridial metabolic pathways, such as an incomplete TCA cycle. A thermodynamic analysis was used to determine the physiological conditions under which proposed pathways (e.g., reverse partial TCA cycle and reverse arginine biosynthesis pathway) are feasible. The reconstructed metabolic network was used to create a genome-scale model that correctly characterized the butyrate kinase knock-out and the asolventogenic M5 pSOL1 megaplasmid degenerate strains. Systematic gene knock-out simulations were performed to identify a set of genes encoding clostridial enzymes essential for growth in silico.     

Construction of a Genome-Scale Metabolic Model of Arthrospira platensis NIES-39 and Metabolic Design for Cyanobacterial Bioproduction

Arthrospira (Spirulina) platensis is a promising feedstock and host strain for bioproduction because of its high accumulation of glycogen and superior characteristics for industrial production. Metabolic simulation using a genome-scale metabolic model and flux balance analysis is a powerful method that can be used to design metabolic engineering strategies for the improvement of target molecule production. In this study, we constructed a genome-scale metabolic model of A. platensis NIES-39 including 746 metabolic reactions and 673 metabolites, and developed novel strategies to improve the production of valuable metabolites, such as glycogen and ethanol. The simulation results obtained using the metabolic model showed high consistency with experimental results for growth rates under several trophic conditions and growth capabilities on various organic substrates. The metabolic model was further applied to design a metabolic network to improve the autotrophic production of glycogen and ethanol. Decreased flux of reactions related to the TCA cycle and phosphoenolpyruvate reaction were found to improve glycogen production. Furthermore, in silico knockout simulation indicated that deletion of genes related to the respiratory chain, such as NAD(P)H dehydrogenase and cytochrome-c oxidase, could enhance ethanol production by using ammonium as a nitrogen source.     

Likelihood-Based Gene Annotations for Gap Filling and Quality Assessment in Genome-Scale Metabolic Models

Genome-scale metabolic models provide a powerful means to harness information from genomes to deepen biological insights. With exponentially increasing sequencing capacity, there is an enormous need for automated reconstruction techniques that can provide more accurate models in a short time frame. Current methods for automated metabolic network reconstruction rely on gene and reaction annotations to build draft metabolic networks and algorithms to fill gaps in these networks. However, automated reconstruction is hampered by database inconsistencies, incorrect annotations, and gap filling largely without considering genomic information. Here we develop an approach for applying genomic information to predict alternative functions for genes and estimate their likelihoods from sequence homology. We show that computed likelihood values were significantly higher for annotations found in manually curated metabolic networks than those that were not. We then apply these alternative functional predictions to estimate reaction likelihoods, which are used in a new gap filling approach called likelihood-based gap filling to predict more genomically consistent solutions. To validate the likelihood-based gap filling approach, we applied it to models where essential pathways were removed, finding that likelihood-based gap filling identified more biologically relevant solutions than parsimony-based gap filling approaches. We also demonstrate that models gap filled using likelihood-based gap filling provide greater coverage and genomic consistency with metabolic gene functions compared to parsimony-based approaches. Interestingly, despite these findings, we found that likelihoods did not significantly affect consistency of gap filled models with Biolog and knockout lethality data. This indicates that the phenotype data alone cannot necessarily be used to discriminate between alternative solutions for gap filling and therefore, that the use of other information is necessary to obtain a more accurate network. All described workflows are implemented as part of the DOE Systems Biology Knowledgebase (KBase) and are publicly available via API or command-line web interface.     

Systematic assignment of thermodynamic constraints in metabolic network models

Background  

The availability of genome sequences for many organisms enabled the reconstruction of several genome-scale metabolic network models. Currently, significant efforts are put into the automated reconstruction of such models. For this, several computational tools have been developed that particularly assist in identifying and compiling the organism-specific lists of metabolic reactions. In contrast, the last step of the model reconstruction process, which is the definition of the thermodynamic constraints in terms of reaction directionalities, still needs to be done manually. No computational method exists that allows for an automated and systematic assignment of reaction directions in genome-scale models.     

Genome-wide metabolic network reconstruction of the picoalga Ostreococcus

The green picoalga Ostreococcus is emerging as a simple plant model organism, and two species, O. lucimarinus and O. tauri, have now been sequenced and annotated manually. To evaluate the completeness of the metabolic annotation of both species, metabolic networks of O. lucimarinus and O. tauri were reconstructed from the KEGG database, thermodynamically constrained, elementally balanced, and functionally evaluated. The draft networks contained extensive gaps and, in the case of O. tauri, no biomass components could be produced due to an incomplete Calvin cycle. To find and remove gaps from the networks, an extensive reference biochemical reaction database was assembled using a stepwise approach that minimized the inclusion of microbial reactions. Gaps were then removed from both Ostreococcus networks using two existing gap-filling methodologies. In the first method, a bottom-up approach, a minimal list of reactions was added to each model to enable the production of all metabolites included in our biomass equation. In the second method, a top-down approach, all reactions in the reference database were added to the target networks and subsequently trimmed away based on the sequence alignment scores of identified orthologues. Because current gap-filling methods do not produce unique solutions, a quality metric that includes a weighting for phylogenetic distance and sequence similarity was developed to distinguish between gap-filling results automatically. The draft O. lucimarinus and O. tauri networks required the addition of 56 and 70 reactions, respectively, in order to produce the same biomass precursor metabolites that were produced by our plant reference database.     

Genome-scale metabolic network analysis of the opportunistic pathogen Pseudomonas aeruginosa PAO1

Pseudomonas aeruginosa is a major life-threatening opportunistic pathogen that commonly infects immunocompromised patients. This bacterium owes its success as a pathogen largely to its metabolic versatility and flexibility. A thorough understanding of P. aeruginosa's metabolism is thus pivotal for the design of effective intervention strategies. Here we aim to provide, through systems analysis, a basis for the characterization of the genome-scale properties of this pathogen's versatile metabolic network. To this end, we reconstructed a genome-scale metabolic network of Pseudomonas aeruginosa PAO1. This reconstruction accounts for 1,056 genes (19% of the genome), 1,030 proteins, and 883 reactions. Flux balance analysis was used to identify key features of P. aeruginosa metabolism, such as growth yield, under defined conditions and with defined knowledge gaps within the network. BIOLOG substrate oxidation data were used in model expansion, and a genome-scale transposon knockout set was compared against in silico knockout predictions to validate the model. Ultimately, this genome-scale model provides a basic modeling framework with which to explore the metabolism of P. aeruginosa in the context of its environmental and genetic constraints, thereby contributing to a more thorough understanding of the genotype-phenotype relationships in this resourceful and dangerous pathogen.     

Reconstruction and In Silico Analysis of Metabolic Network for an Oleaginous Yeast,Yarrowia lipolytica

With the emergence of energy scarcity, the use of renewable energy sources such as biodiesel is becoming increasingly necessary. Recently, many researchers have focused their minds on Yarrowia lipolytica, a model oleaginous yeast, which can be employed to accumulate large amounts of lipids that could be further converted to biodiesel. In order to understand the metabolic characteristics of Y. lipolytica at a systems level and to examine the potential for enhanced lipid production, a genome-scale compartmentalized metabolic network was reconstructed based on a combination of genome annotation and the detailed biochemical knowledge from multiple databases such as KEGG, ENZYME and BIGG. The information about protein and reaction associations of all the organisms in KEGG and Expasy-ENZYME database was arranged into an EXCEL file that can then be regarded as a new useful database to generate other reconstructions. The generated model iYL619_PCP accounts for 619 genes, 843 metabolites and 1,142 reactions including 236 transport reactions, 125 exchange reactions and 13 spontaneous reactions. The in silico model successfully predicted the minimal media and the growing abilities on different substrates. With flux balance analysis, single gene knockouts were also simulated to predict the essential genes and partially essential genes. In addition, flux variability analysis was applied to design new mutant strains that will redirect fluxes through the network and may enhance the production of lipid. This genome-scale metabolic model of Y. lipolytica can facilitate system-level metabolic analysis as well as strain development for improving the production of biodiesels and other valuable products by Y. lipolytica and other closely related oleaginous yeasts.     

Identification of functional differences in metabolic networks using comparative genomics and constraint-based models

Genome-scale network reconstructions are useful tools for understanding cellular metabolism, and comparisons of such reconstructions can provide insight into metabolic differences between organisms. Recent efforts toward comparing genome-scale models have focused primarily on aligning metabolic networks at the reaction level and then looking at differences and similarities in reaction and gene content. However, these reaction comparison approaches are time-consuming and do not identify the effect network differences have on the functional states of the network. We have developed a bilevel mixed-integer programming approach, CONGA, to identify functional differences between metabolic networks by comparing network reconstructions aligned at the gene level. We first identify orthologous genes across two reconstructions and then use CONGA to identify conditions under which differences in gene content give rise to differences in metabolic capabilities. By seeking genes whose deletion in one or both models disproportionately changes flux through a selected reaction (e.g., growth or by-product secretion) in one model over another, we are able to identify structural metabolic network differences enabling unique metabolic capabilities. Using CONGA, we explore functional differences between two metabolic reconstructions of Escherichia coli and identify a set of reactions responsible for chemical production differences between the two models. We also use this approach to aid in the development of a genome-scale model of Synechococcus sp. PCC 7002. Finally, we propose potential antimicrobial targets in Mycobacterium tuberculosis and Staphylococcus aureus based on differences in their metabolic capabilities. Through these examples, we demonstrate that a gene-centric approach to comparing metabolic networks allows for a rapid comparison of metabolic models at a functional level. Using CONGA, we can identify differences in reaction and gene content which give rise to different functional predictions. Because CONGA provides a general framework, it can be applied to find functional differences across models and biological systems beyond those presented here.     

Systemic metabolic reactions are obtained by singular value decomposition of genome-scale stoichiometric matrices

Genome-scale metabolic networks can be reconstructed. The systemic biochemical properties of these networks can now be studied. Here, genome-scale reconstructed metabolic networks were analysed using singular value decomposition (SVD). All the individual biochemical conversions contained in a reconstructed metabolic network are described by a stoichiometric matrix (S). SVD of S led to the definition of the underlying modes that characterize the overall biochemical conversions that take place in a network and rank-ordered their importance. The modes were shown to correspond to systemic biochemical reactions and they could be used to identify the groups and clusters of individual biochemical reactions that drive them. Comparative analysis of the Escherichia coli, Haemophilus influenzae, and Helicobacter pylori genome-scale metabolic networks showed that the four dominant modes in all three networks correspond to: (1) the conversion of ATP to ADP, (2) redox metabolism of NADP, (3) proton-motive force, and (4) inorganic phosphate metabolism. The sets of individual metabolic reactions deriving these systemic conversions, however, differed among the three organisms. Thus, we can now define systemic metabolic reactions, or eigen-reactions, for the study of systems biology of metabolism and have a basis for comparing the overall properties of genome-specific metabolic networks.     

Pan-cancer analysis of the metabolic reaction network

Metabolic reprogramming is considered a hallmark of malignant transformation. However, it is not clear whether the network of metabolic reactions expressed by cancers of different origin differ from each other or from normal human tissues. In this study, we reconstructed functional and connected genome-scale metabolic models for 917 primary tumor samples across 13 types based on the probability of expression for 3765 reference metabolic genes in the sample. This network-centric approach revealed that tumor metabolic networks are largely similar in terms of accounted reactions, despite diversity in the expression of the associated genes. On average, each network contained 4721 reactions, of which 74% were core reactions (present in >95% of all models). Whilst 99.3% of the core reactions were classified as housekeeping also in normal tissues, we identified reactions catalyzed by ARG2, RHAG, SLC6 and SLC16 family gene members, and PTGS1 and PTGS2 as core exclusively in cancer. These findings were subsequently replicated in an independent validation set of 3388 genome-scale metabolic models. The remaining 26% of the reactions were contextual reactions. Their inclusion was dependent in one case (GLS2) on the absence of TP53 mutations and in 94.6% of cases on differences in cancer types. This dependency largely resembled differences in expression patterns in the corresponding normal tissues, with some exceptions like the presence of the NANP-encoded reaction in tumors not from the female reproductive system or of the SLC5A9-encoded reaction in kidney-pancreatic-colorectal tumors. In conclusion, tumors expressed a metabolic network virtually overlapping the matched normal tissues, raising the possibility that metabolic reprogramming simply reflects cancer cell plasticity to adapt to varying conditions thanks to redundancy and complexity of the underlying metabolic networks. At the same time, the here uncovered exceptions represent a resource to identify selective liabilities of tumor metabolism.     

Genome-scale metabolic model for Lactococcus lactis MG1363 and its application to the analysis of flavor formation

Lactococcus lactis subsp. cremoris MG1363 is a paradigm strain for lactococci used in industrial dairy fermentations. However, despite of its importance for process development, no genome-scale metabolic model has been reported thus far. Moreover, current models for other lactococci only focus on growth and sugar degradation. A metabolic model that includes nitrogen metabolism and flavor-forming pathways is instrumental for the understanding and designing new industrial applications of these lactic acid bacteria. A genome-scale, constraint-based model of the metabolism and transport in L. lactis MG1363, accounting for 518 genes, 754 reactions, and 650 metabolites, was developed and experimentally validated. Fifty-nine reactions are directly or indirectly involved in flavor formation. Flux Balance Analysis and Flux Variability Analysis were used to investigate flux distributions within the whole metabolic network. Anaerobic carbon-limited continuous cultures were used for estimating the energetic parameters. A thorough model-driven analysis showing a highly flexible nitrogen metabolism, e.g., branched-chain amino acid catabolism which coupled with the redox balance, is pivotal for the prediction of the formation of different flavor compounds. Furthermore, the model predicted the formation of volatile sulfur compounds as a result of the fermentation. These products were subsequently identified in the experimental fermentations carried out. Thus, the genome-scale metabolic model couples the carbon and nitrogen metabolism in L. lactis MG1363 with complete known catabolic pathways leading to flavor formation. The model provided valuable insights into the metabolic networks underlying flavor formation and has the potential to contribute to new developments in dairy industries and cheese-flavor research.