Genome-scale metabolic models predict diet-and lifestyle-driven shifts of ecological interactions in the gut microbiome

Genome-scale metabolic models predict diet-and lifestyle-driven shifts of ecological interactions in the gut microbiome | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Genome-scale metabolic models predict diet-and lifestyle-driven shifts of ecological interactions in the gut microbiome View ORCID Profile Georgios Marinos , View ORCID Profile Karlis Arturs Moors , View ORCID Profile Malte Rühlemann , View ORCID Profile Silvio Waschina , View ORCID Profile Wolfgang Lieb , View ORCID Profile Andre Franke , View ORCID Profile Matthias Laudes , View ORCID Profile Mathieu Groussin , View ORCID Profile Mathilde Poyet , View ORCID Profile Christoph Kaleta , View ORCID Profile A. Samer Kadibalban doi: https://doi.org/10.1101/2025.09.23.678088 Georgios Marinos 1 Research Group Medical Systems Biology, Institute of Experimental Medicine, University Hospital Schleswig-Holstein Campus Kiel, Kiel University , Kiel, Schleswig-Holstein, Germany 2 CAU Innovation GmbH, Kiel University , Kiel, Schleswig-Holstein, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Georgios Marinos Karlis Arturs Moors 1 Research Group Medical Systems Biology, Institute of Experimental Medicine, University Hospital Schleswig-Holstein Campus Kiel, Kiel University , Kiel, Schleswig-Holstein, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Karlis Arturs Moors Malte Rühlemann 4 Institute of Clinical Molecular Biology, University Hospital Schleswig-Holstein Campus Kiel, Kiel University , Kiel, Schleswig-Holstein, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Malte Rühlemann Silvio Waschina 3 Nutriinformatics, Institute of Human Nutrition and Food Science, Kiel University , Kiel, Schleswig-Holstein, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Silvio Waschina Wolfgang Lieb 6 Institute of Epidemiology, Kiel, Kiel University , Kiel, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Wolfgang Lieb Andre Franke 4 Institute of Clinical Molecular Biology, University Hospital Schleswig-Holstein Campus Kiel, Kiel University , Kiel, Schleswig-Holstein, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Andre Franke Matthias Laudes 7 Institute of Diabetes and Clinical Metabolic Research, University Hospital Schleswig-Holstein Campus Kiel , Kiel, Schleswig-Holstein, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Matthias Laudes Mathieu Groussin 4 Institute of Clinical Molecular Biology, University Hospital Schleswig-Holstein Campus Kiel, Kiel University , Kiel, Schleswig-Holstein, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Mathieu Groussin Mathilde Poyet 5 Institute of Experimental Medicine, University Hospital Schleswig-Holstein Campus Kiel, Kiel University , Kiel, Schleswig-Holstein, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Mathilde Poyet Christoph Kaleta 1 Research Group Medical Systems Biology, Institute of Experimental Medicine, University Hospital Schleswig-Holstein Campus Kiel, Kiel University , Kiel, Schleswig-Holstein, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Christoph Kaleta For correspondence: c.kaleta{at}iem.uni-kiel.de s.kadibalban{at}iem.uni-kiel.de A. Samer Kadibalban 1 Research Group Medical Systems Biology, Institute of Experimental Medicine, University Hospital Schleswig-Holstein Campus Kiel, Kiel University , Kiel, Schleswig-Holstein, Germany 8 Current address: Institute of Clinical Chemistry, University Hospital Schleswig-Holstein Campus Kiel, Kiel University, Kiel , Schleswig-Holstein, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for A. Samer Kadibalban For correspondence: c.kaleta{at}iem.uni-kiel.de s.kadibalban{at}iem.uni-kiel.de Abstract Full Text Info/History Metrics Supplementary material Data/Code Preview PDF 1. Abstract Microbiomes and their host environments form complex, interconnected ecosystems. The microbial species within a microbiome, on the one hand, compete for resources, while on the other hand, they exchange vital metabolites to support their survival. These interactions are influenced by the microbial genetic repertoire, environmental conditions, and the availability of nutrients. We developed EcoGS ( http://www.github.com/KaletaLab/EcoGS ), a metabolic modelling tool designed to predict ecological interactions between pairs of microbes. Applying EcoGS to the microbiomes of two distinct human cohorts revealed a shift from collaborative to exploitative ecological interactions associated with higher dietary intake of simple sugars (glucose and fructose), in diabetic individuals and those living in industrialised lifestyles. On the other hand, the consumption of Cobalamin (vitamin B12), phylloquinone (vitamin K1) and biotin (vitamin B7), among other compounds, was associated with increased collaboration in the gut microbiome. We conclude that the abundance of simple sugars as an energy source reduces the necessity for microbes to cooperate, thereby increasing competition and hostility among microbiome members. Moreover, our study proposes multiple compounds, such as urate, deoxyadenosine, deoxyguanosine, and hypoxanthine, for in vitro validation tests as dietary interventions that have the potential to restore the ecological balance amongst the community. EcoGS serves as a valuable tool for exploring microbiome dynamics and their connections to environmental changes and disease. 3. Introduction The interactions among microbes, as well as between microbes and their host, are complex, yet understanding these relationships is critical for elucidating how the microbiome influences human health. The gut microbiome plays a key role in a variety of diseases, including inflammatory conditions such as Inflammatory Bowel Disease, where current therapies primarily target inflammation rather than directly modulating the microbial community [ 1 ]. It also contributes to metabolic disorders, including obesity and diabetes, and influences host immune function and behaviour through the gut-brain axis [ 2 ]. Like all microbial communities, the trillions of microbes in the human gut do not act in isolation; they form intricate ecological networks that mediate cooperation, competition, and metabolic exchange. Although each individual’s microbiome is unique, external factors such as lifestyle and environment significantly shape community composition and function [ 3 ]. These considerations highlight the importance of investigating microbiome structure and function, as well as the diversity and stability of microbial ecological networks, using advanced computational tools. Ecological theory considers that microorganisms interact with each other and their environment, resulting in different growth profiles for individual microorganisms in comparison to growth in isolation. Those interactions can be exploitative (amensalism, competition, and antagonism), collaborative (commensalism and mutualism) or neutral [ 4 ]. To study those interactions, a basic approach is to model pairs of two species without taking into account the effect of other species. To this end, the co-growth capabilities of two species are directly associated with their ability to compete or collaborate for resources. Specifically, their co-growth rates should be less than their single individual growth rates in case of competition, more in case of collaboration, and the same in case of absence of interaction [ 5 ]. To understand interactions within the microbiome, it is important to study the co-occurrence of the involved microbes. Metabolically dissimilar species are more likely to co-occur, as they do not compete for the same resources. Furthermore, metabolic competition is positively associated with phylogenetic relatedness [ 6 , 7 ]. However, metabolic cooperation can still be predicted. Specifically, cases of close cooperation loops among species, where the products of one species are a nutritional source for other species, have been predicted in communities [ 5 ]. Interestingly, cooperation can lead to the stability of the communities in case of nutritional perturbations, while competition ensures stability in case invasive species occur [ 6 ]. While the above-mentioned ecological concepts appear theoretically straightforward, inferring them for the gut microbiome is challenging. Importantly, this is because a vast number of microbes cannot be cultivated [ 8 ], and the physicochemical environment of the gut that strongly influences microbiome structure is highly variable [ 9 ]. Moreover, the human gut microbiome is influenced by human lifestyle and environmental factors, as urbanisation has put pressure on species that were favoured by the pre-agricultural way of life [ 3 ]. To this end, a seminal review on the effect of nutritional groups (e.g., amino acids, fats, carbohydrates) and dietary styles (e.g., Western diet) on the abundance of species was published [ 10 ]. For instance, the lack of microbiota-accessible carbohydrates in Western-style diets is a reason for species extinction in the gut, which otherwise could produce short-chain fatty acids, leading to detrimental effects, such as dysbiosis, for the host [ 11 ]. There is a need for tools that allow categorising ecological interactions within the microbiome as a prerequisite for a targeted manipulation of microbiome structure and function [ 12 ]. One key approach that allows for computationally predicting ecological interactions within microbial communities is constraint-based modelling. This methodology builds on genome-scale metabolic reconstructions of the metabolic networks of microbial species that can be derived using automatic tools, like gapseq [ 13 ]. Therefore, apart from the genome of the species, nutritional cues, which serve as a proxy for lifestyle and environment, can also be taken into account. Such models can be used for simulating growth profiles of individual microbial species based on linear programming approaches (e.g. flux balance analysis (FBA) [ 14 ]). Furthermore, it is possible to simulate complex microbial communities using approaches such as community FBA. This technique merges multiple metabolic models into one, which can then be used in FBA simulations. The MicrobiomeGS2 [ 15 ] is such a modelling software, which has been used in reconstructing personalised community microbial models of humans [ 16 ] and nematodes [ 17 ]. To predict ecological interactions within microbial communities, we present a user-friendly implementation of our approach that predicts ecological interactions between pairs of species within a given community and nutritional input. The software is based on MicrobiomeGS2 and accepts SBML metabolic models that are produced by gapseq [ 13 ] and loaded with sybil [ 18 ] as input. It is implemented as an R package available on GitHub ( www.github.com/maringos/EcoGS ). In our study, we used this tool to predict the ecological interactions of gut microbial species in the microbiomes of a German cohort and to investigate the stability of the proposed interactions by inducing perturbations to the nutritional input of the models. Lastly, we expanded our investigation by comparing and contrasting the prevalence of the ecological relations among individuals from industrialised and non-industrialised settings. 4. Methods 4.1. Cohort Data In this study, data from two independent cohorts were analysed. The FoCus, BSP, and SPC cohorts, here referred to collectively as the Focus cohort (PopGen Biobank, Schleswig-Holstein, Germany), included 2,396 individuals and encompassed comprehensive dietary data, 16S rRNA gene sequences of the gut microbiome, as well as clinical data. Metadata variables included age, gender, body mass index (BMI), smoking status, diabetes, inflammatory bowel disease (IBD), and other relevant health parameters. The cohort consisted of 1,329 females and 1,067 males, with 274 individuals having diabetes (Supplementary figure S1). The data of the 108 individuals diagnosed with IBD were filtered out before the downstream analysis, as IBD has a substantial influence on the gut microbiome structure and dynamics [ 19 ]. The 16S rRNA sequences of the gut microbiome of the FoCus cohort were mapped against a reference database of 822 human gut microbial genomes derived from the AGORA resource [ 20 ] using USEARCH version v10 [ 21 ]. This mapping identified 409 microbial strains with varying abundances across the cohort. The second cohort in this study comprised 1,229 participants as a part of the Global Microbiome Conservancy (GMbC) [ 22 ]. This cohort consisted of 789 individuals from non-industrialised communities and 440 from industrialised communities (Supplementary figure S2). For further analysis, we used both the metagenomic data generated as well as whole-genome assemblies. The metadata included age, gender, and body mass index (BMI). 4.2. Creating and simulating the community models gapseq (version 1.2) [ 13 ] was used to create genome-scale metabolic models for both the reference (AGORA) microbial genomes and the metagenomic bins of the GMbC cohort, utilising a standardised North German diet [ 16 ]. Thereafter, the biomass reaction of the metabolic models was optimised based on flux balance analysis using the R [ 23 ] package sybil [ 18 ], and the achieved baseline growth rate was recorded. Subsequently, the models were joined in pairs using the R package MicrobiomeGS2. After joining each pair of models, and to simulate the community growth, the objective coefficients of each sub-model were set to one; the objective coefficients of other reactions were set to 0. In the next step, following a published approach, we set the following coupling constraints: c = 400 mmol.g -1 .hr -1 as proposed originally [ 24 ] and a more relaxed coupling u = 1e-6 mmol.g -1 .hr -1 Then we performed flux balance analysis, as originally designed in MicrobiomeGS2. We explicitly recorded the exchange reactions whose fluxes predict the exchange of compounds among the sub-models when one flux is positive and the other is negative, respectively. To ensure the quality of the results, we recorded for each case if both submodels could grow (cut-off 1e -6 ) and if each optimisation process was successful. Both the simulations of the stand-alone models and of the joined models were conducted using the academic version of the linear solver IBM ILOG CPLEX through its R interface, cplexAPI [ 25 ], and using the Sybil method hybbaropt. Both simulations run in parallel using multi-core computing based on PSOCK clusters in R. 4.3. Inference of the ecological interactions For each of the 409 microbial species found in participants in the FoCus cohort and the 483 vspecies found in participants of the GMbC cohort, we compared the predicted alone-growth rate (AG) with its co-growth rate (CG) when joined with each of the other microbial species belonging to the same cohort. A change in co-growth rate was considered significant if it differed from the alone-growth rate by at least 1% ( Figure 1 and Table 1 ). Based on these comparisons, we constructed a symmetrical matrix representing the inferred ecological interactions between species pairs for each of the two cohorts, encompassing a total of 83,436 pairs for the FoCus microbes (Supplementary Table S1) and 116,403 pairs for the GMbC microbes (Supplementary Table S1). The number of pairs can be calculated as follows Download figure Open in new tab Figure 1: Diagram representation of the EcoGS pipeline: Red boxes represent metabolic model data, blue boxes represent predictions extracted from the models, purple diamond shapes represent operations and functions included in the package, continuous arrows represent the data flow between each step, and dashed arrows represent the feedback of candidate intervention data for predicting their influence. View this table: View inline View popup Table 1: Definition of the ecological interactions for two species, a and b Next, we categorised each microbial species pair into one of the possible six ecological interactions ( Table 1 ). Thereafter, we counted the number of pairs belonging to each ecological interaction in each microbial community (4,236 communities in the two cohorts combined) and weighed those counts by the abundance of the species in each pair. Let the abundances of species S1, S2, S3, S4, S5 , etc. be denoted as A1, A2, A3, A4, A5 , etc. For a given ecological interaction type E , the total number of species pairs exhibiting this interaction in the community is NE, represented as: (S1, S2), (S1, S3), (S3, S5), etc. The weighted estimate of this ecological interaction, W NE , can be calculated in two ways: W1 NE takes into account the probability of two species encountering each other in the community as the multiplication of their abundances. W2 NE takes into account that the limiting factor in the metabolic exchanges between the two species is the minimum abundance of the two. EcoGS allows the user to choose the weighing method. Here, we opted to use the second approach (W2 NE ), as it showed stronger signals in the results. To address the inherent codependency in compositional data, a logarithmic transformation of the ratio between pairs of variables is often applied [ 26 ]. In this study, the proportions of different ecological interactions are compositional data. In the context of the six microbial ecological interactions, this approach yielded 15 log-transformed ecological interaction ratios (EIRs), such as log(Antagonism/Competition) , log(Commensalism/Mutualism) , and so forth. The log-transformed proportional data of the ecological interaction ratios per microbial community (with pseudoanonymised IDs) are in supplementary table S1 for the FoCus cohort and in supplementary table S1 for the GmBC cohort. 4.4 Phylogenetic inference of the GMbC species Phylogenetic trees were reconstructed using SGB representative genomes with the identify, align, and infer functions of GTDB-Tk v2.3.2 and GTDB Release R07-RS207 [ 27 , 28 ]. The bacterial and archaeal trees were rooted with representatives of the Patescibacteria and Nanoarchaeota, respectively, and joined at the root to yield a combined phylogenetic tree encompassing both domains. 4.5 Dietary interventions To predict the effects of dietary interventions on ecological interactions within the gut microbiome of the FoCus cohort, simulations were performed for the supplementation of 105 individual nutrients (metabolites). Each simulation involved modifying the microbial diet in gapseq models by adjusting the lower bounds of the exchange reaction for the respective metabolite. The supplementation amount for each nutrient was set to two times the value of the standardised North German diet (Supplementary Table S1). For each intervention, the ecological interaction prediction pipeline was applied to microbial pairs independently, generating 105 ecological interaction matrices with the same dimensions and structure as the original pre-intervention matrix. Thereafter, post-intervention ecological interaction ratios per participant were calculated. 4.6 Statistical analysis For the FoCus cohort, linear models (LMs) were employed to assess associations between the ratios of ecological interactions (EIRs) and four key clinical variables: age, body mass index (BMI), gender, and smoking behaviour (measured as cigarettes per day), as well as with diabetes. The model structure was specified as: Additionally, EIRs were tested for associations with individual dietary metabolites, which were inferred from dietary questionnaires. Each metabolite was analysed separately, adjusting for the same four clinical covariates: Similarly, for the GMbC cohort, linear models were applied to examine the associations between EIRs and age, BMI, gender, and lifestyle (classified as industrialised or non-industrialised): Following the metabolic supplementation simulations, the effect of each supplementation on the EIRs was evaluated by comparing the EIRs before and after supplementation using a paired Wilcoxon signed-rank test. Finally, the phylogenetic distances between each pair of microbes in the GMbC cohort were compared across the six ecological interaction categories using pairwise Wilcoxon signed-rank tests. False discovery rate (FDR) [ 29 ] correction was applied to the resulting P -values to account for multiple testing after each statistical test. Adjusted P -values less than α = 0.05 were considered statistically significant. 4.7 Software The community model simulations were conducted in a high-performance computing environment in an R environment (version 4.4.1) [ 23 ]. Specifically, the following R packages were used: MicrobiomeGS2 [ 15 ] version 0.2.0, cplexAPI [ 25 ] version 1.4.0 (along with the IBM ILOG CPLEX Optimisation Studio / Academic Initiative version 22.10), sybil [ 18 ] version 2.2.1, doParallel [ 30 ] version 1.0.17, stringr version 1.5.1, [ 31 ] lattice version 0.22-6, Matrix version 1.7-1, iterators version 1.0.1 4, data.table version 1.17.8, and foreach[ 32 ] version 1.5.2. All statistical analyses were conducted with R (version 4.5.0) within the RStudio environment [ 33 ]. Plots were created with the ggplot2 R package (version 3.5.2) [ 34 ] and R base plotting functions. The results of all statistical tests were determined to be significant below a threshold of α=0.05. P-values were adjusted for multiple testing using FDR [ 29 ]. 5. Results 5.1. Ecological interaction and metabolic exchange among microbial communities and across phylogeny While ecological interaction categories (amensalism, competition, antagonism, commensalism, mutualism, and neutralism) address the relationship between the interacting species, it is also noteworthy to examine the interaction effect on each model (microbial species) independently (harmful, beneficial, and neutral). The ecological interactions among the gut microbial communities of both cohorts have considerably higher proportions and higher variances of exploitative ecological interactions in comparison to the collaborative and neutral interactions ( Figures 2 . A and 2B). Furthermore, species pairs with exploitative interactions tend to have higher phylogenetic distances with a bimodality, suggesting that very closely related species also tend to have exploitative interactions, while collaborative and neutral interactions tend to have lower to intermediate phylogenetic distances as observed in the GMbC cohort ( Figure 2C ). In order to evaluate the differences in the ecological interactions between different microbial genera, the 18 most abundant genera in the gut communities of the FoCus cohort participants were selected. Download figure Open in new tab Figure 2: Ecological interactions among microbial communities. A. The frequencies of the six types of ecological interactions among the gut microbiome of the FoCus cohort; the x-axis is ordered by the median of the ecological interaction frequencies across all communities. B. Similar to A. for the GmbC cohort microbial communities, the colours correspond to those in A. C. A violin plot representing the distribution of the phylogenetic distances between pairs of microbes among the six different types of ecological interactions for microbes found in the GMbC cohort participants. The black line in the middle of each box represents the median value. The violins are sorted by median on the x-axis. The colours correspond to those in A. and B. D. Bar plots depicting the interaction outcome for the 18 most abundant genera in the FoCus cohort. The top part shows the outcome of interacting with species from different genera. The bottom part shows the outcome of interacting with species belonging to the same genus. The genera are ordered on the X-axis to correspond to the order in E. E. Bar plots depicting the ecological interaction proportions for the 18 most abundant genera in the FoCus cohort. The top part shows the ecological interactions between microbes of the corresponding genus and different genera. The bottom part shows the ecological interactions between species belonging to the same genus. The genera are ordered on the X-axis in correspondence to their frequency of amensalistic interaction with other genera. The interaction effect of each model was estimated within each genus and between genera. The former takes into account the interactions among microbial species belonging to each of the genera, and the latter between the species of each of the genera and the species of all other genera. ( Figure 2D ). Similarly, the proportion of the ecological interactions (considering the interaction effect on both microbes in the pair) was estimated for microbial pairs within each genus and between genera ( Figure 2E ). The proportions of interaction effects, on each model, vary between the different genera; the interactions among species from the same genus produce more neutral and less beneficial effects than interactions with species from different genera. The ecological interaction categories also vary between different genera. Higher antagonism, lower competition and higher commensalism are observed in interactions between species belonging to different genera in comparison to interactions within the same genus. Not only does Escherichia have the highest frequency of antagonistic interactions in comparison to other genera, but also, antagonism comprises the highest proportion of all types of ecological interactions between Escherichia and other genera. Moreover, those antagonistic interactions are mostly harmful to Escherichia ( Figures 2D and 2E and supplementary table S2). Campylobacter, Clostridium , and Eubacterium, on the other hand, have the highest proportions of collaborative interactions with other genera and within the same genus. Whereas Fusobacterium , Prevotella , Ruminococcus , and Eubacterium have relatively higher competition against other genera ( Figure 2E and Supplementary Table S2). On the other hand, Bacteroides species showed a highly amensalistic trend between genera and a competitive trend within the genus. Staphylococcus exhibits a very high proportion of amensalistic interactions within the same genus. Furthermore, we observed a positive association between the number of predicted metabolic exchanges among microbial pairs in the GMbC cohort and their phylogenetic distance; this relationship remained significant even when considering only amino acid exchanges ( Figures 3A and 3B ). In contrast, a similar pattern was not evident in the FoCus cohort. When comparing exchange frequencies within the same genus versus across different genera—after normalising by the number of interacting pairs in each category—we found no strong differences for either total metabolic exchanges or amino acid-specific exchanges. The only notable exception was Listeria and Prevotella , which exhibited substantially more amino acid exchanges within their respective genera than with other genera. Download figure Open in new tab Figure 3: The number of metabolic exchanges between microbial pairs across phylogeny and genera. A. The number of predicted exchanged metabolites between microbial pairs on the Y-axis, across three binned categories of phylogenetic distance between the microbes in each pair on the X-axis. B. Same as A., but only counting the number of amino acid exchanges between the microbes in the pairs. C. A bar plot of the normalised rates of predicted metabolic exchanges between microbes in pairs within the same genus in orange and those belonging to different genera in green. D. Same as C., but only counting the normalised rates of amino acid exchanges between the microbes in the pairs. 5.2. Associations between phenotypic and dietary data and the ecological interaction ratios The linear regression models revealed significant associations between ecological interaction ratios (EIRs) and phenotypical data from both cohorts (e.g. a positive association between diabetes and the proportion antagonism/mutualism and antagonism/commensalism, Figure 4A ). Such a positive estimate reflects an increase in the numerator (antagonism) relative to the denominators (mutualism and commensalism). Download figure Open in new tab Figure 4: Associations between phenotypic and dietary data with ecological relation ratios. A. Associations between metadata of the FoCus cohort participants and the ecological interaction ratios in their gut microbiomes. Each dot represents one significant association resulting from (LM) between four phenotypic (metadata) on the Y-axis and the ecological interaction ratios on the X-axis (log transformed). The colour represents the directionality of the association, blue for positive and red for negative associations. The size of the dots depicts the LM estimate (coefficient of correlation). The interaction ratios are divided into three panels to distinguish between three types of interaction ratios: Exploitative-exploitative, Exploitative-collaborative, and Collaborative-collaborative. To enhance readability, we grouped the compounds into categories (W for water, V for vitamin, SA for sugar alcohol, S for sugar, P for polyamines, N for Nitrogen-containing compounds, I for ions, F for fatty acids, and AA for amino acids). B. Associations between dietary elements from the diet of the FoCus cohort (y-axis) and the estimated ecological relation ratios of their gut microbiomes (x-axis). The colour and size of the dots and the two panels are as in A (there are only two panels because there are no significant associations with the collaborative-collaborative type of interaction ratios). C. Associations between metadata of the GMbC cohort participants (x-axis) and the ecological interaction ratios in their gut microbiomes (y-axis). The colour and size of the dots and the two panels are as in A. Similarly, a negative estimate reflects a decrease in the numerator relative to the denominator (e.g. lifestyle vs. antagonism/neutralism, Figure 4B ). In the FoCus cohort, antagonism proportion relative to commensalism and mutualism is significantly increased in the gut microbiome of female participants in comparison to males (P values in supplementary table S2). However, we don’t observe the same effect in the GMbC cohort ( Figures 4A and 3B ). Smoking, on the other hand, was associated with higher neutral interactions relative to collaborative and exploitative ones and higher antagonism relative to mutualism ( Figure 4A ). A positive association between BMI and the proportion of antagonism relative to commensalism was observed in the FoCus cohort. However, in the GMbC cohort, we only observe a decrease in exploitative interactions relative to neutralism in association with BMI ( Figures 4A and 4B ). Moreover, not only is there a relative increase in antagonism relative to collaborative interactions associated with diabetes, BMI, smoking and in females, but antagonism is also increased relative to the two other exploitative interactions (amensalism and competition) in association with the same four phenotypical data types. Furthermore, we observed an increase in amensalism, competition and antagonism at the expense of collaborative ecological interactions in individuals living in industrialised communities in comparison to individuals living non-industrialised lifestyles in the GMbC cohort. However, the proportions of exploitative and collaborative interactions relative to neutral ones are significantly lower in industrialised populations. ( Figure 4B ) Furthermore, we tested for associations between EIRs and different nutrients from the diet of the FoCus cohort. 32 of the tested dietary elements showed a negative association with at least one exploitative relative to collaborative ecological interaction types, including vitamins, ions, fatty acids, nucleotides, polyamines, as well as the amino acid Tryptophan. The majority of those dietary elements had significant associations with increased commensalism. The strongest associations were observed between Cobalamin (vitamin B12), phylloquinone (vitamin K1) and biotin (vitamin B7) with reduced exploitative relative to collaborative interaction ratios. On the other hand, we observed a significant increase in all three exploitative interactions relative to commensalism associated with higher dietary glucose and fructose intake, as well as an increase in antagonism relative to mutualism in association with glucose intake ( Figure 4C ). 5.3. Dietary interventions to alter the ecological interaction distributions along gut microbiomes To identify dietary components that could alter the ecological interaction ratios and restore the observed shifts associated with diseases such as diabetes, we assessed the impact of individual nutrients on predicted interaction ratios (see Methods). Among the tested metabolites, 19 showed a significant effect on these ratios. Supplementation with certain compounds, including the fatty acid octadecanoate, the nucleotide cytidine, seven purines (deoxyadenosine, guanosine, xanthine, urate, deoxyguanosine, hypoxanthine, and adenosine), and two vitamins (thiamin B1 and riboflavin B2), was associated with an increase in collaborative interactions (mutualism and commensalism) relative to exploitative interactions (antagonism and competition). Notably, supplementation with D-glucose and D-fructose increased collaborative interactions, whereas L-tryptophan supplementation increased exploitative interactions ( Figures 4C and 5 ; see Discussion). Download figure Open in new tab Figure 5: The effect of dietary metabolic intervention (Y-axis) on the ecological relation ratios (X-axis), the dots represent the significant difference between the interaction ratios with and without the intervention, detected with the Wilcoxon signed-rank test. The colour of the dots represents the direction of the change caused by the intervention, red for a decrease in the interaction ratio and blue for an increase. The size of the dots shows the fold change caused by an intervention, depicted as the median of all relative changes among all communities. The two panels discriminate between exploitative-collaborative and exploitative-exploitative interaction ratios. Within the exploitative category, several interventions altered internal ratios, with most increasing amensalism relative to antagonism and competition, and in some cases, increasing antagonism relative to competition. Notably, L-tryptophan produced an opposite pattern. Additional shifts were observed between exploitative and neutral interactions. Some metabolites promoted amensalism over neutralism, some promoted neutralism over competition, and only spermidine increased neutralism relative to all three exploitative interaction types ( Figure 5 ). 6. Discussion In this study, we introduce EcoGS , an open-source R-based computational tool designed to infer ecological interactions among microbial species from genome-scale metabolic models. By simulating the growth of each species both alone and together with another, EcoGS predicts interaction types: Exploitative interactions (amensalism, antagonism, and competition), collaborative interactions (commensalism and mutualism) and neutral interactions and summarises them at the community level. This framework offers a scalable and user-friendly approach for exploring microbial ecology in complex communities and is freely available as an R package ( www.github.com/KaletaLab/EcoGS ). Applying EcoGS to gut microbiomes from two distinct human cohorts, the FoCus cohort from Germany and the GmbC cohort encompassing both industrialised and non-industrialised populations, yielded key insights into how host-related metabolic factors, lifestyle, and diet shape microbial interaction networks. Notably, in both cohorts, we observe higher proportions of exploitative interactions than collaborative ones ( Figures 2A and 2B ). This pattern is in line with the previously observed frequent co-occurrence among competing species, suggesting that shared habitat filtering drives microbiome structure [ 7 ]. It also aligns with controlled experiments showing that competitive and amensal interactions dominate in synthetic human gut communities (comprising ∼68 % of all pairwise interactions, vs. ∼5 % cooperative ones) [ 12 ], as well as with recent data showing predominance of negative interactions in natural gut isolates [ 35 ]. The phylogenetic distance of interacting species revealed more exploitative interactions between distant species ( Figure 2C ). However, we also observe a bimodal distribution of pairwise distances among microbes engaged in exploitative interactions, suggesting that those interactions also occur among closely related and distantly related taxa ( Figure 2C ). In addition, a positive correlation between the number of metabolic exchanges and phylogenetic distance ( Figures 3A and 3B ) supports the idea that closer species are metabolically similar, share similar enzyme systems, carbon source preferences, and oxygen requirements. Due to this niche overlap and evolutionary constraints on metabolic complementation, they tend to engage in less frequent metabolite sharing [ 36 ]. In the FoCus cohort, we observe an increase in antagonism relative to commensalism in individuals with a higher body mass index (BMI) and in individuals with diabetes, as well as an increase in antagonism relative to mutualism in diabetic individuals. This shift away from collaborative or neutral interactions may reflect a disrupted or less cooperative microbiome in metabolically compromised states, potentially reinforcing disease-associated dysbiosis. However, in the GMbC cohort, there is only an increase in neutralism over other types of interactions in association with BMI. This could be due to the structure of the cohort, which includes individuals from different lifestyles. With this, our study extends findings from a previous study [ 37 ] across independent populations. In that previous study, it was found that elevated blood glucose levels and the presence of diabetes were associated with a marked increase in exploitative microbial interactions, particularly antagonism. Moreover, in the GMbC cohort, microbiomes from participants with industrialised lifestyles displayed a dual trend: A rise in exploitative interactions over collaborative ones, alongside an increase in neutral interactions. This suggests that industrialisation not only tilts microbial communities toward more competitive dynamics but also introduces ecological stasis or tolerance, where species coexist without significant interaction, a possible reflection of ecological simplification driven by uniform diets, reduced microbial diversity, or environmental exposures associated with industrialisation [ 38 ]. Together, these findings indicate that both metabolic disease and industrialisation converge on a common ecological footprint: reduced cooperation and increased exploitation within the gut microbial community. Both diabetes and industrialisation often imply higher availability of simple carbohydrates, highlighting a potential mechanism by which modern diets contribute to microbiome instability and drive metabolic dysregulation, as the availability of simple, rapidly metabolised energy sources reduces the need for microbial cooperation. In the human gut, it is suggested that species have distinct roles, from hydrolysing complex carbohydrates to primary and secondary fermenters that produce short-chain fatty acids and species that use hydrogen for their metabolism [ 39 ]. However, simple sugars like glucose are universally possible to assimilate and energetically cheap to metabolise, which makes them an evolutionarily preferred carbon source when available [ 40 ]. When such substrates are abundant, most community members can independently access energy without relying on complementary metabolic pathways provided by other microbes. This leads to a breakdown in mutualistic and commensal interactions, as cross-feeding and metabolic interdependence become less advantageous. As a result, community dynamics shift toward individualism, with species competing more directly for the same resources, thereby fostering exploitative interactions and reducing ecological cohesion. A consistent driver across these interaction shifts is their association with nutrient intake, particularly simple sugars. In the FoCus cohort, consumption of glucose and fructose is associated with an increase in exploitative interactions, suggesting that these rapidly metabolised substrates may amplify microbial competition or antagonism for shared resources. In contrast, consumption of vitamins, nitrogen-containing compounds, polyamines, and tryptophan promoted collaborative interactions, particularly commensalism. This is in line with the notion that those compounds are expensive to build up or must be consumed by the environment as public goods [ 39 ]. Together, our findings highlight a consistent ecological signal across disease, lifestyle, and diet, which is why dietary modulation may offer a strategy for restoring cooperative microbial interactions in dysbiotic or industrialised microbiomes. However, this potential remains constrained by modelling limitations. The intervention analysis highlights that targeted nutrient supplementation can modulate the ecological interaction structure of the gut microbiome, but these effects are not always aligned with observational associations from dietary intake. For example, while higher habitual intake of simple sugars (glucose and fructose) was linked to an increase in exploitative interactions in cohort analyses, supplementation of these same sugars in silico increased collaborative interactions. This apparent contradiction likely reflects fundamental differences between dietary patterns and supplementation scenarios in a controlled metabolic model. In the real world, sustained high intake of simple sugars may reduce resource diversity and increase competition for easily accessible substrates, promoting exploitative dynamics. By contrast, supplementation in a model context can create an abundance of a readily metabolizable carbon source for both species, reducing competitive pressure. This “dose” dependency in turn increases their predicted growth rates and would be later interpreted as an increase in collaboration. Vitamins and purines consistently promoted collaborative dynamics in silico, supporting the view that micronutrients and nucleotides act as public goods within microbial communities [ 41 , 42 ]. These compounds often require specialised biosynthetic pathways, so supplementation may relieve metabolic bottlenecks and encourage positive-sum exchanges, reinforcing cooperative interactions. Besides, we observe that nitrogen-containing compounds (cytidine, deoxyadenosine, guanosine, xanthine, urate, deoxyguanosine, hypoxanthine, and adenosine) are boosting collaborative interactions. This is particularly interesting if we consider that the colon, except towards its end, is usually a nitrogen-depleted environment [ 39 ] Collectively, these findings emphasise that nutrient availability strongly governs microbial interaction patterns, but the effect direction can depend on abundance context, metabolic accessibility, and whether resources are universally utilisable or pathway-specific. This suggests that dietary interventions aimed at restoring microbial cooperation must consider not only nutrient type but also systemic context and metabolic redundancy within the community. Currently, EcoGS models only pairwise interactions, which may fail to capture emergent behaviours in more complex microbial consortia. Although higher-order modelling is theoretically possible, the combinatorial explosion, e.g., ∼89 million triplet interactions for all strains in the AGORA gut collection, renders it computationally infeasible with current approaches. Moreover, precision diet modelling is limited by the need to re-run simulations for each specific diet, which is computationally intensive. Finally, the reliance on gap-filling algorithms during genome-scale reconstruction, via gapseq, may reduce the number of unique metabolic exchanges, biasing interaction classifications toward competition. In summary, EcoGS represents a valuable tool for investigating microbial ecology at scale. Its ability to uncover consistent interaction shifts across human populations and dietary patterns underscores the microbiome’s role as a mediator and driver of host-environment interactions. Future developments, such as scalable multi-species modelling, dynamic nutrient flux integration, and tighter coupling with empirical validation, will further enhance its utility in both ecological theory and microbiome-based intervention design. 8. Declarations 8.1. Ethics approval and consent to participate Permission to utilise the FoCus cohort data, which included ethics approval, was obtained by submitting a structured online application (available at https://www.uksh.de/P2N/ ) to the Kiel 2.0 Network (Reference Number: 2024-009). Permission to utilise the GMbC cohort data was obtained by the co-authors, GMbC consortium leaders, and science & ethics advisory board members Mathieu Groussin and Mathilde Poyet. Research & ethics approvals were obtained from the MIT IRB (protocol #1612797956) and from the Ethics Commission of the Medical Faculty of Kiel University (Studie D 511/24). Permits were also obtained in each sampled country prior to the start of sample collection, from the following ethics committees: ● Cameroon: Comité National d’Ethique de la Recherche pour la Santé Humaine, protocol #2017/05/901/CE/CNERSH/SP; ● Canada: Nunavut Research Institute, protocol #0205217N-M; ● Central African Republic: Comité d’Éthique et Scientifique de la Faculté des Sciences de la Santé de l’Université de Bangui, No. 17/UB/FACSS/CES20; ● Finland: Coordinating Ethics Committee of Helsinki and Uusimaa Hospital District, protocol #1527/2017; ● Ghana: Cape Coast Teaching Hospital Ethical Review Committee, protocol #CCTHERC/RS/EC/2016/3 and Committee on Human Research, Publication and Ethics of the Komfo Anokye Teaching Hospital, protocol #CHRPE/AP/398/18; ● Malaysia: Universiti Malaya Medical Research Ethics Committee, MREC ID No.: 2018219-6033; ● Nigeria: National Health Research Ethics Committee of Nigeria, protocol #NHREC/01/01/2007-29/04/2018. ● Rwanda: National Ethics Committee, protocol IRB 00001497 of IORG0001100 ● Senegal: Comité National d’Ethique pour la recherche en Santé, No.: 0000073 MSAS/DPRS/CNERS ● Tanzania: National Institute for Medical Research, protocol #NIMR/HQ/R.8a/Vol. IX/2657; ● Thailand: The Human Research Ethics Committee of the Faculty of Medicine of Thammasat University No.1, Certificate of Approval 143/2019 ● United States of America: Chief Dull Knife College, protocol #FWA00020985; 8.2. Availability of data and materials The statistical analysis and plotting R scripts can be found under: https://doi.org/10.6084/m9.figshare.30149005 9. Supplementary material Download figure Open in new tab Figure S1: The distribution of FoCus cohort descriptive phenotypic data Download figure Open in new tab Figure S2: The distribution of GmBC cohort descriptive phenotypic data Supplementary table S1: https://docs.google.com/spreadsheets/d/1IXCp2RYRWFko31_K3P7fXAb4CcNPRfxyJunhAKgZyx4/edit?usp=sharing 8.3. Acknowledgements We thank the following funding organisations ‒ Excellence Cluster for Precision Medicine in Chronic Inflammation (PMI) ‒ The German Research Foundation within the collaborative research centre “Origin and Function of Metaorganisms,” CRC1182 ‒ Research Group miTarget ‒ iTREAT ‒ DFG project ExoMod We thank Dr Johannes Zimmermann and Dr Robin Koch for the discussions and support Funder Information Declared Excellence Cluster for Precision Medicine in Chronic Inflammation (PMI) The German Research Foundation within the collaborative research centre ‘Origin and Function of Metaorganisms,‘ , CRC1182 Research Group miTarget iTREAT DFG project ExoMod Footnotes https://figshare.com/articles/software/R_scripts_for_statistics_and_plotting_that_were_used_to_generate_the_Pvalues_and_plots_published_in_the_EcoGS_manuscript/30149005 http://www.github.com/KaletaLab/EcoGS 7. List of abbreviations EcoGS The presented tool to predict the Eco logical interactions based on G enome- s cale metabolic models GMbC The G lobal M icro b iome C onservancy EIR In this document refers to the E cological I nteraction R atios FDR F alse D iscovery R ate, a method for correcting for multiple testing BMI B ody M ass I ndex, calculated by dividing the weight by the height squared 10. References 1. ↵ Iliev ID , Ananthakrishnan AN , Guo C-J . Microbiota in inflammatory bowel disease: mechanisms of disease and therapeutic opportunities . Nat Rev Microbiol 2025 ; 23 : 509 – 524 . doi: 10.1038/s41579-025-01163-0 OpenUrl CrossRef PubMed 2. ↵ Thomas S et al. The Host Microbiome Regulates and Maintains Human Health: A Primer and Perspective for Non-Microbiologists . Cancer Res 2017 ; 77 : 1783 – 1812 . doi: 10.1158/0008-5472.CAN-16-2929 OpenUrl Abstract / FREE Full Text 3. ↵ Mancabelli L et al. Meta-analysis of the human gut microbiome from urbanized and pre-agricultural populations . Environ Microbiol 2017 ; 19 : 1379 – 1390 . doi: 10.1111/1462-2920.13692 OpenUrl CrossRef 4. ↵ Zélé F et al. Ecology and evolution of facilitation among symbionts . Nat Commun 2018 ; 9 :4869. doi: 10.1038/s41467-018-06779-w OpenUrl CrossRef PubMed 5. ↵ Freilich S et al. Competitive and cooperative metabolic interactions in bacterial communities . Nat Commun 2011 ; 2 : 589 . doi: 10.1038/ncomms1597 OpenUrl CrossRef PubMed 6. ↵ Machado D et al. Polarization of microbial communities between competitive and cooperative metabolism . Nat Ecol Evol 2021 ; 5 : 195 – 203 . doi: 10.1038/s41559-020-01353-4 OpenUrl CrossRef 7. ↵ Levy R , Borenstein E . Metabolic modeling of species interaction in the human microbiome elucidates community-level assembly rules . Proc Natl Acad Sci U S A 2013 ; 110 : 12804 – 12809 . doi: 10.1073/pnas.1300926110 OpenUrl Abstract / FREE Full Text 8. ↵ Kuziel GA , Rakoff-Nahoum S. The gut microbiome . Curr Biol CB 2022 ; 32 :R257–R264. doi: 10.1016/j.cub.2022.02.023 OpenUrl CrossRef PubMed 9. ↵ Procházková N et al. Gut physiology and environment explain variations in human gut microbiome composition and metabolism . Nat Microbiol 2024 ; 9 : 3210 – 3225 . doi: 10.1038/s41564-024-01856-x OpenUrl CrossRef PubMed 10. ↵ Singh RK et al. Influence of diet on the gut microbiome and implications for human health . J Transl Med 2017 ; 15 : 73 . doi: 10.1186/s12967-017-1175-y OpenUrl CrossRef PubMed 11. ↵ Sonnenburg ED , Sonnenburg JL . Starving our microbial self: the deleterious consequences of a diet deficient in microbiota-accessible carbohydrates . Cell Metab 2014 ; 20 : 779 – 786 . doi: 10.1016/j.cmet.2014.07.003 OpenUrl CrossRef PubMed 12. ↵ Coyte KZ , Rakoff-Nahoum S . Understanding Competition and Cooperation within the Mammalian Gut Microbiome . Curr Biol CB 2019 ; 29 : R538 – R544 . doi: 10.1016/j.cub.2019.04.017 OpenUrl CrossRef PubMed 13. ↵ Zimmermann J , Kaleta C , Waschina S. gapseq: informed prediction of bacterial metabolic pathways and reconstruction of accurate metabolic models . Genome Biol 2021 ; 22 : 81 . doi: 10.1186/s13059-021-02295-1 OpenUrl CrossRef PubMed 14. ↵ Orth JD , Thiele I , Palsson BØ . What is flux balance analysis? Nat Biotechnol 2010 ; 28 : 245 – 248 . doi: 10.1038/nbt.1614 OpenUrl CrossRef PubMed Web of Science 15. ↵ . Waschina S. Analysis and simulation of genome-& ecosystem-scale microbial metabolism [Internet] . Available from: www.github.com/Waschina/MicrobiomeGS2 . 16. ↵ Pryor R et al. Host-Microbe-Drug-Nutrient Screen Identifies Bacterial Effectors of Metformin Therapy . Cell 2019 ; 178 : 1299 – 1312 .e29. doi: 10.1016/j.cell.2019.08.003 OpenUrl CrossRef PubMed 17. ↵ Marinos G et al. Metabolic model predictions enable targeted microbiome manipulation through precision prebiotics . Microbiol Spectr 2024 ; 12 : e0114423 . doi: 10.1128/spectrum.01144-23 OpenUrl CrossRef 18. ↵ Gelius-Dietrich G et al. Sybil--efficient constraint-based modelling in R . BMC Syst Biol 2013 ; 7 : 125 . doi: 10.1186/1752-0509-7-125 OpenUrl CrossRef PubMed 19. ↵ Taubenheim J et al. Metabolic modeling reveals a multi-level deregulation of host-microbiome metabolic networks in IBD . Nat Commun 2025 ; 16 :5120. doi: 10.1038/s41467-025-60233-2 OpenUrl CrossRef 20. ↵ Magnúsdóttir S et al. Generation of genome-scale metabolic reconstructions for 773 members of the human gut microbiota . Nat Biotechnol 2017 ; 35 : 81 – 89 . doi: 10.1038/nbt.3703 OpenUrl CrossRef PubMed 21. ↵ Edgar RC . Search and clustering orders of magnitude faster than BLAST . Bioinformatics 2010 ; 26 : 2460 – 2461 . doi: 10.1093/bioinformatics/btq461 OpenUrl CrossRef PubMed Web of Science 22. ↵ Groussin M et al. Elevated rates of horizontal gene transfer in the industrialized human microbiome . Cell 2021 ; 184 : 2053 – 2067 .e18. doi: 10.1016/j.cell.2021.02.052 OpenUrl CrossRef PubMed 23. ↵ R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. 24. ↵ Heinken A et al. Systems-level characterization of a host-microbe metabolic symbiosis in the mammalian gut . Gut Microbes 2013 ; 4 : 28 – 40 . doi: 10.4161/gmic.22370 OpenUrl CrossRef PubMed 25. ↵ Gelius-Dietrich G. cplexAPI: R Interface to C API of IBM ILOG CPLEX. 26. ↵ Greenacre M. Compositional Data Analysis . Annu Rev Stat Its Appl 2021 ; 8 : 271 – 299 . doi: 10.1146/annurev-statistics-042720-124436 OpenUrl CrossRef 27. ↵ Chaumeil P-A et al. GTDB-Tk: a toolkit to classify genomes with the Genome Taxonomy Database . Bioinformatics 2020 ; 36 : 1925 – 1927 . doi: 10.1093/bioinformatics/btz848 OpenUrl CrossRef PubMed 28. ↵ Parks DH et al. GTDB: an ongoing census of bacterial and archaeal diversity through a phylogenetically consistent, rank normalized and complete genome-based taxonomy . Nucleic Acids Res 2022 ; 50 : D785 – D794 . doi: 10.1093/nar/gkab776 OpenUrl CrossRef PubMed 29. ↵ Benjamini Y , Hochberg Y . Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing . J R Stat Soc Ser B Methodol 1995 ; 57 : 289 – 300 . doi: 10.1111/j.2517-6161.1995.tb02031.x OpenUrl CrossRef PubMed 30. ↵ Microsoft Corporation, Weston S. doParallel: Foreach Parallel Adaptor for the ‘parallel’ Package. 31. ↵ Sarkar D . Lattice: Multivariate Data Visualization with R. Springer , New York , 2008 . 32. ↵ . Microsoft Corporation, Weston S. foreach: Provides Foreach Looping Construct. 33. ↵ Posit team ( 2023 ). RStudio: Integrated Development Environment for R. Posit Software, PBC, Boston, MA. URL http://www.posit.co/ . 34. ↵ Wickham H ( 2016 ). ggplot2: Elegant Graphics for Data Analysis . Springer-Verlag New York . ISBN 978-3-319-24277-4, https://ggplot2.tidyverse.org . 35. ↵ Zhu J et al. Systematic pairwise co-cultures uncover predominant negative interactions among human gut bacteria . Microbiome 2025 ; 13 : 161 . doi: 10.1186/s40168-025-02156-0 OpenUrl CrossRef 36. ↵ Mataigne V et al. Multi-genome metabolic modeling predicts functional inter-dependencies in the Arabidopsis root microbiome . Microbiome 2022 ; 10 : 217 . doi: 10.1186/s40168-022-01383-z OpenUrl CrossRef PubMed 37. ↵ Kadibalban AS et al. Metabolic modelling reveals increased autonomy and antagonism in type 2 diabetic gut microbiota . Mol Syst Biol 2025 ; 21 : 720 – 731 . doi: 10.1038/s44320-025-00100-w OpenUrl CrossRef PubMed 38. ↵ Sonnenburg JL , Sonnenburg ED. Vulnerability of the industrialized microbiota . Science 2019 . doi: 10.1126/science.aaw9255 OpenUrl Abstract / FREE Full Text 39. ↵ Culp EJ , Goodman AL . Cross-feeding in the gut microbiome: Ecology and mechanisms . Cell Host Microbe 2023 ; 31 : 485 – 499 . doi: 10.1016/j.chom.2023.03.016 OpenUrl CrossRef PubMed 40. ↵ Aidelberg G et al. Hierarchy of non-glucose sugars in Escherichia coli . BMC Syst Biol 2014 ; 8 : 133 . doi: 10.1186/s12918-014-0133-z OpenUrl CrossRef PubMed 41. ↵ Sharma V et al. B-Vitamin Sharing Promotes Stability of Gut Microbial Communities . Front Microbiol 2019 ; 10 :1485. doi: 10.3389/fmicb.2019.01485 OpenUrl CrossRef PubMed 42. ↵ Chuang Y-C et al. A purine salvage bottleneck leads to bacterial adenine cross-feeding . bioRxiv 2023 ;2023.10.17.562681. doi: 10.1101/2023.10.17.562681 OpenUrl Abstract / FREE Full Text View the discussion thread. Back to top Previous Next Posted September 25, 2025. Download PDF Supplementary Material Data/Code Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Genome-scale metabolic models predict diet-and lifestyle-driven shifts of ecological interactions in the gut microbiome Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Genome-scale metabolic models predict diet-and lifestyle-driven shifts of ecological interactions in the gut microbiome Georgios Marinos , Karlis Arturs Moors , Malte Rühlemann , Silvio Waschina , Wolfgang Lieb , Andre Franke , Matthias Laudes , Mathieu Groussin , Mathilde Poyet , Christoph Kaleta , A. Samer Kadibalban bioRxiv 2025.09.23.678088; doi: https://doi.org/10.1101/2025.09.23.678088 Share This Article: Copy Citation Tools Genome-scale metabolic models predict diet-and lifestyle-driven shifts of ecological interactions in the gut microbiome Georgios Marinos , Karlis Arturs Moors , Malte Rühlemann , Silvio Waschina , Wolfgang Lieb , Andre Franke , Matthias Laudes , Mathieu Groussin , Mathilde Poyet , Christoph Kaleta , A. Samer Kadibalban bioRxiv 2025.09.23.678088; doi: https://doi.org/10.1101/2025.09.23.678088 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Systems Biology Subject Areas All Articles Animal Behavior and Cognition (7621) Biochemistry (17645) Bioengineering (13867) Bioinformatics (41872) Biophysics (21416) Cancer Biology (18549) Cell Biology (25443) Clinical Trials (138) Developmental Biology (13360) Ecology (19866) Epidemiology (2067) Evolutionary Biology (24289) Genetics (15587) Genomics (22470) Immunology (17706) Microbiology (40314) Molecular Biology (17142) Neuroscience (88456) Paleontology (666) Pathology (2826) Pharmacology and Toxicology (4815) Physiology (7634) Plant Biology (15111) Scientific Communication and Education (2042) Synthetic Biology (4285) Systems Biology (9812) Zoology (2268)