Genome-scale metabolic models reveal determinants of phenotypic differences in non-Saccharomyces yeasts.

BACKGROUND: Use of alternative non-Saccharomyces yeasts in wine and beer brewing has gained more attention the recent years. This is both due to the desire to obtain a wider variety of flavours in the product and to reduce the final alcohol content. Given the metabolic differences between the yeast species, we wanted to account for some of the differences by using in silico models. RESULTS: We created and studied genome-scale metabolic models of five different non-Saccharomyces species using an automated processes. These were: Metschnikowia pulcherrima, Lachancea thermotolerans, Hanseniaspora osmophila, Torulaspora delbrueckii and Kluyveromyces lactis. Using the models, we predicted that M. pulcherrima, when compared to the other species, conducts more respiration and thus produces less fermentation products, a finding which agrees with experimental data. Complex I of the electron transport chain was to be present in M. pulcherrima, but absent in the others. The predicted importance of Complex I was diminished when we incorporated constraints on the amount of enzymatic protein, as this shifts the metabolism towards fermentation. CONCLUSIONS: Our results suggest that Complex I in the electron transport chain is a key differentiator between Metschnikowia pulcherrima and the other yeasts considered. Yet, more annotations and experimental data have the potential to improve model quality in order to increase fidelity and confidence in these results. Further experiments should be conducted to confirm the in vivo effect of Complex I in M. pulcherrima and its respiratory metabolism.

Predictive evolution of metabolic phenotypes using model-designed environments.

Adaptive evolution under controlled laboratory conditions has been highly effective in selecting organisms with beneficial phenotypes such as stress tolerance. The evolution route is particularly attractive when the organisms are either difficult to engineer or the genetic basis of the phenotype is complex. However, many desired traits, like metabolite secretion, have been inaccessible to adaptive selection due to their trade-off with cell growth. Here, we utilize genome-scale metabolic models to design nutrient environments for selecting lineages with enhanced metabolite secretion. To overcome the growth-secretion trade-off, we identify environments wherein growth becomes correlated with a secondary trait termed tacking trait. The latter is selected to be coupled with the desired trait in the application environment where the trait manifestation is required. Thus, adaptive evolution in the model-designed selection environment and subsequent return to the application environment is predicted to enhance the desired trait. We experimentally validate this strategy by evolving Saccharomyces cerevisiae for increased secretion of aroma compounds, and confirm the predicted flux-rerouting using genomic, transcriptomic, and proteomic analyses. Overall, model-designed selection environments open new opportunities for predictive evolution.

csdR, an R package for differential co-expression analysis.

BACKGROUND: Differential co-expression network analysis has become an important tool to gain understanding of biological phenotypes and diseases. The CSD algorithm is a method to generate differential co-expression networks by comparing gene co-expressions from two different conditions. Each of the gene pairs is assigned conserved (C), specific (S) and differentiated (D) scores based on the co-expression of the gene pair between the two conditions. The result of the procedure is a network where the nodes are genes and the links are the gene pairs with the highest C-, S-, and D-scores. However, the existing CSD-implementations suffer from poor computational performance, difficult user procedures and lack of documentation. RESULTS: We created the R-package csdR aimed at reaching good performance together with ease of use, sufficient documentation, and with the ability to play well with other tools for data analysis. csdR was benchmarked on a realistic dataset with 20,645 genes. After verifying that the chosen number of iterations gave sufficient robustness, we tested the performance against the two existing CSD implementations. csdR was superior in performance to one of the implementations, whereas the other did not run. Our implementation can utilize multiple processing cores. However, we were unable to achieve more than [Formula: see text]2.7 parallel speedup with saturation reached at about 10 cores. CONCLUSION: The results suggest that csdR is a useful tool for differential co-expression analysis and is able to generate robust results within a workday on datasets of realistic sizes when run on a workstation or compute server.

Genome-scale metabolic modelling when changes in environmental conditions affect biomass composition.

Genome-scale metabolic modeling is an important tool in the study of metabolism by enhancing the collation of knowledge, interpretation of data, and prediction of metabolic capabilities. A frequent assumption in the use of genome-scale models is that the in vivo organism is evolved for optimal growth, where growth is represented by flux through a biomass objective function (BOF). While the specific composition of the BOF is crucial, its formulation is often inherited from similar organisms due to the experimental challenges associated with its proper determination. A cell's macro-molecular composition is not fixed and it responds to changes in environmental conditions. As a consequence, initiatives for the high-fidelity determination of cellular biomass composition have been launched. Thus, there is a need for a mathematical and computational framework capable of using multiple measurements of cellular biomass composition in different environments. Here, we propose two different computational approaches for directly addressing this challenge: Biomass Trade-off Weighting (BTW) and Higher-dimensional-plane InterPolation (HIP). In lieu of experimental data on biomass composition-variation in response to changing nutrient environment, we assess the properties of BTW and HIP using three hypothetical, yet biologically plausible, BOFs for the Escherichia coli genome-scale metabolic model iML1515. We find that the BTW and HIP formulations have a significant impact on model performance and phenotypes. Furthermore, the BTW method generates larger growth rates in all environments when compared to HIP. Using acetate secretion and the respiratory quotient as proxies for phenotypic changes, we find marked differences between the methods as HIP generates BOFs more similar to a reference BOF than BTW. We conclude that the presented methods constitute a conceptual step in developing genome-scale metabolic modelling approaches capable of addressing the inherent dependence of cellular biomass composition on nutrient environments.

Automatic reconstruction of metabolic pathways from identified biosynthetic gene clusters.

BACKGROUND: A wide range of bioactive compounds is produced by enzymes and enzymatic complexes encoded in biosynthetic gene clusters (BGCs). These BGCs can be identified and functionally annotated based on their DNA sequence. Candidates for further research and development may be prioritized based on properties such as their functional annotation, (dis)similarity to known BGCs, and bioactivity assays. Production of the target compound in the native strain is often not achievable, rendering heterologous expression in an optimized host strain as a promising alternative. Genome-scale metabolic models are frequently used to guide strain development, but large-scale incorporation and testing of heterologous production of complex natural products in this framework is hampered by the amount of manual work required to translate annotated BGCs to metabolic pathways. To this end, we have developed a pipeline for an automated reconstruction of BGC associated metabolic pathways responsible for the synthesis of non-ribosomal peptides and polyketides, two of the dominant classes of bioactive compounds. RESULTS: The developed pipeline correctly predicts 72.8% of the metabolic reactions in a detailed evaluation of 8 different BGCs comprising 228 functional domains. By introducing the reconstructed pathways into a genome-scale metabolic model we demonstrate that this level of accuracy is sufficient to make reliable in silico predictions with respect to production rate and gene knockout targets. Furthermore, we apply the pipeline to a large BGC database and reconstruct 943 metabolic pathways. We identify 17 enzymatic reactions using high-throughput assessment of potential knockout targets for increasing the production of any of the associated compounds. However, the targets only provide a relative increase of up to 6% compared to wild-type production rates. CONCLUSION: With this pipeline we pave the way for an extended use of genome-scale metabolic models in strain design of heterologous expression hosts. In this context, we identified generic knockout targets for the increased production of heterologous compounds. However, as the predicted increase is minor for any of the single-reaction knockout targets, these results indicate that more sophisticated strain-engineering strategies are necessary for the development of efficient BGC expression hosts.

ErrorTracer: an algorithm for identifying the origins of inconsistencies in genome-scale metabolic models.

MOTIVATION: The number and complexity of genome-scale metabolic models is steadily increasing, empowered by automated model-generation algorithms. The quality control of the models, however, has always remained a significant challenge, the most fundamental being reactions incapable of carrying flux. Numerous automated gap-filling algorithms try to address this problem, but can rarely resolve all of a model's inconsistencies. The need for fast inconsistency checking algorithms has also been emphasized with the recent community push for automated model-validation before model publication. Previously, we wrote a graphical software to allow the modeller to solve the remaining errors manually. Nevertheless, model size and complexity remained a hindrance to efficiently tracking origins of inconsistency. RESULTS: We developed the ErrorTracer algorithm in order to address the shortcomings of existing approaches: ErrorTracer searches for inconsistencies, classifies them and identifies their origins. The algorithm is ∼2 orders of magnitude faster than current community standard methods, using only seconds even for large-scale models. This allows for interactive exploration in direct combination with model visualization, markedly simplifying the whole error-identification and correction work flow. AVAILABILITY AND IMPLEMENTATION: Windows and Linux executables and source code are available under the EPL 2.0 Licence at https://github.com/TheAngryFox/ModelExplorer and https://www.ntnu.edu/almaaslab/downloads. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

High-quality genome-scale metabolic model of Aurantiochytrium sp. T66.

The long-chain, ω-3 polyunsaturated fatty acids (PUFAs) (e.g., eicosapentaenoic acid [EPA] and docosahexaenoic acid [DHA]), are essential for humans and animals, including marine fish species. Presently, the primary source of these PUFAs is fish oils. As the global production of fish oils appears to be reaching its limits, alternative sources of high-quality ω-3 PUFAs is paramount to support the growing aquaculture industry. Thraustochytrids are a group of heterotrophic protists with the capability to synthesize and accrue large amounts of DHA. Thus, the thraustochytrids are prime candidates to solve the increasing demand for ω-3 PUFAs using microbial cell factories. However, a systems-level understanding of their metabolic shift from cellular growth into lipid accumulation is, to a large extent, unclear. Here, we reconstructed a high-quality genome-scale metabolic model of the thraustochytrid Aurantiochytrium sp. T66 termed iVS1191. Through iterative rounds of model refinement and extensive manual curation, we significantly enhanced the metabolic scope and coverage of the reconstruction from that of previously published models, making considerable improvements with stoichiometric consistency, metabolic connectivity, and model annotations. We show that iVS1191 is highly consistent with experimental growth data, reproducing in vivo growth phenotypes as well as specific growth rates on minimal carbon media. The availability of iVS1191 provides a solid framework for further developing our understanding of T66's metabolic properties, as well as exploring metabolic engineering and process-optimization strategies in silico for increased ω-3 PUFA production.

Containing pandemics through targeted testing of households.

BACKGROUND: While invasive social distancing measures have proven efficient to control the spread of pandemics failing wide-scale deployment of vaccines, they carry vast societal costs. The development of a diagnostic methodology for identifying COVID-19 infection through simple testing was a reality only a few weeks after the novel virus was officially announced. Thus, we were interested in exploring the ability of regular testing of non-symptomatic people to reduce cases and thereby offer a non-pharmaceutical tool for controlling the spread of a pandemic. METHODS: We developed a data-driven individual-based epidemiological network model in order to investigate epidemic countermeasures. This models is based on high-resolution demographic data for each municipality in Norway, and each person in the model is subject to Susceptible-Exposed-Infectious-Recovered (SEIR) dynamics. The model was calibrated against hospitalization data in Oslo, Norway, a city with a population of 700k which we have used as the simulations focus. RESULTS: Finding that large households function as hubs for the propagation of COVID-19, we assess the intervention efficiency of targeted pooled household testing (TPHT) repeatedly. For an outbreak with reproductive number R=1.4, we find that weekly TPHT of the 25% largest households brings R below unity. For the case of R=1.2, our results suggest that TPHT with the largest 25% of households every three days in an urban area is as effective as a lockdown in curbing the outbreak. Our investigations of different disease parameters suggest that these results are markedly improved for disease variants that more easily infect young people, and when compliance with self-isolation rules is less than perfect among suspected symptomatic cases. These results are quite robust to changes in the testing frequency, city size, and the household-size distribution. Our results are robust even with only 50% of households willing to participate in TPHT, provided the total number of tests stay unchanged. CONCLUSIONS: Pooled and targeted household testing appears to be a powerful non-pharmaceutical alternative to more invasive social-distancing and lock-down measures as a localized early response to contain epidemic outbreaks.

Identification of key proteins involved in stickleback environmental adaptation with system-level analysis.

Using abundance measurements of 1,490 proteins from four separate populations of three-spined sticklebacks, we implemented a system-level approach to correlate proteome dynamics with environmental salinity and temperature and the fish's population and morphotype. We identified robust and accurate fingerprints that classify environmental salinity, temperature, morphotype, and the population sample origin, observing that proteins with specific functions are enriched in these fingerprints. Highly apparent functions represented in all fingerprints include ion transport, proteostasis, growth, and immunity, suggesting that these functions are most diversified in populations inhabiting different environments. Applying a differential network approach, we analyzed the network of protein interactions that differs between populations. Looking at specific population combinations of differential interaction, we identify sets of connected proteins. We find that these sets and their corresponding enriched functions reflect key processes that have diverged between the four populations. Moreover, the extent of divergence, i.e., the number of enriched functions that differ between populations, is highest when all three environmental parameters are different between two populations. Key nodes in the differential interaction network signify functions that are also inherent in the fingerprints, most prominently proteostasis-related functions. However, the differential interaction network also reveals additional functions that have diverged between populations, notably cytoskeletal organization and morphogenesis. The strength of these analyses is that the results are purely data driven. With such an unbiased approach applied on a large proteomic data set, we find the strongest signals given by the data, making it possible to develop more discriminatory and complex biomarkers for specific contexts of interest.

Assessment of weighted topological overlap (wTO) to improve fidelity of gene co-expression networks.

BACKGROUND: For more than a decade, gene expression data sets have been used as basis for the construction of co-expression networks used in systems biology investigations, leading to many important discoveries in a wide range of subjects spanning human disease to evolution and the development of organisms. A commonly encountered challenge in such investigations is first that of detecting, then subsequently removing, spurious correlations (i.e. links) in these networks. While access to a large number of measurements per gene would reduce this problem, often only a small number of measurements are available. The weighted Topological Overlap (wTO) measure, which incorporates information from the shared network-neighborhood of a given gene-pair into a single score, is a metric that is frequently used with the implicit expectation of producing higher-quality networks. However, the actual extent to which wTO improves on the accuracy of a co-expression analysis has not been quantified. RESULTS: Here, we used a large-sample biological data set containing 338 gene-expression measurements per gene as a reference system. From these data, we generated ensembles consisting of 10, 20 and 50 randomly selected measurements to emulate low-quality data sets, finding that the wTO measure consistently generates more robust scores than what results from simple correlation calculations. Furthermore, for the data sets consisting of only 10 and 20 samples per gene, we find that wTO serves as a better predictor of the correlation scores generated from the full data set. However, we find that using wTO as a score for network building substantially alters several topographical aspects of the resulting networks, with no conclusive evidence that the resulting structure is more accurate. Importantly, we find that the much used approach of applying a soft-threshold modifier to link weights prior to computing the wTO substantially decreases the robustness of the resulting wTO network, but increases the predictive power of wTO networks with regards to the reference correlation (soft threshold) network, particularly as the size of the data sets increases. CONCLUSION: Our analysis demonstrates that, in agreement with previous assumptions, the wTO approach is capable of significantly improving the fidelity of co-expression networks, and that this effect is especially evident for cases of low-sample number gene-expression data sets.

ModelExplorer - software for visual inspection and inconsistency correction of genome-scale metabolic reconstructions.

BACKGROUND: Genome-scale metabolic network reconstructions are low level chemical representations of biological organisms. These models allow the system-level investigation of metabolic phenotypes using a variety of computational approaches. The link between a metabolic network model and an organisms' higher-level behaviour is usually found using a constraint-based analysis approach, such as FBA (Flux Balance Analysis). However, the process of model reconstruction rarely proceeds without error. Often, considerable parts of a model cannot carry flux under any condition. This is termed model inconsistency and is caused by faulty topology and/or stoichiometry of the underlying reconstructed network. While there exist several automated gap-filling tools that may solve some of the inconsistencies, much of the work still needs to be carried out manually. The common "linear list" format of writing biochemical reactions makes it difficult to intuit what is at the root of the inconsistent behaviour. Unfortunately, we have frequently observed that model builders do not correct their models past the abilities of automated tools, leaving many widely used models significantly inconsistent. RESULTS: We have developed the software ModelExplorer, which main purpose is to fill this gap by providing an intuitive and visual framework that allows the user to explore and correct inconsistencies in genome-scale metabolic models. The software will automatically visualize metabolic networks as graphs with distinct separation and delineation of cellular compartments. ModelExplorer highlights reactions and species that are unable to carry flux (blocked), with several different consistency checking modes available. Our software also allows the automatic identification of neighbours and production pathways of any species or reaction. Additionally, the user may focus on any chosen inconsistent part of the model on its own. This facilitates a rapid and visual identification of reactions and species responsible for model inconsistencies. Finally, ModelExplorer lets the user freely edit, add or delete model elements, allowing straight-forward correction of discovered issues. CONCLUSION: Overall, ModelExplorer is currently the fastest real-time metabolic network visualization program available. It implements several consistency checking algorithms, which in combination with its set of tracking tools, gives an efficient and systematic model-correction process.

GREM1 is associated with metastasis and predicts poor prognosis in ER-negative breast cancer patients.

BACKGROUND: In breast cancer, activation of bone morphogenetic protein (BMP) signaling and elevated levels of BMP-antagonists have been linked to tumor progression and metastasis. However, the simultaneous upregulation of BMPs and their antagonist, and the fact that both promote tumor aggressiveness seems contradictory and is not fully understood. METHODS: We analyzed the transcriptomes of the metastatic 66cl4 and the non-metastatic 67NR cell lines of the 4T1 mouse mammary tumor model to search for factors that promote metastasis. CRISPR/Cas9 gene editing was used for mechanistic studies in the same cell lines. Furthermore, we analyzed gene expression patterns in human breast cancer biopsies obtained from public datasets to evaluate co-expression and possible relations to clinical outcome. RESULTS: We found that mRNA levels of the BMP-antagonist Grem1, encoding gremlin1, and the ligand Bmp4 were both significantly upregulated in cells and primary tumors of 66cl4 compared to 67NR. Depletion of gremlin1 in 66cl4 could impair metastasis to the lungs in this model. Furthermore, we found that expression of Grem1 correlated with upregulation of several stem cell markers in 66cl4 cells compared to 67NR cells. Both in the mouse model and in patients, expression of GREM1 associated with extracellular matrix organization, and formation, biosynthesis and modification of collagen. Importantly, high expression of GREM1 predicted poor prognosis in estrogen receptor negative breast cancer patients. Analyses of large patient cohorts revealed that amplification of genes encoding BMP-antagonists and elevation of the corresponding transcripts is evident in biopsies from more than half of the patients and much more frequent for the secreted BMP-antagonists than the intracellular inhibitors of SMAD signaling. CONCLUSION: In conclusion, our results show that GREM1 is associated with metastasis and predicts poor prognosis in ER-negative breast cancer patients. Gremlin1 could represent a novel target for therapy.

Automated generation of genome-scale metabolic draft reconstructions based on KEGG.

BACKGROUND: Constraint-based modeling is a widely used and powerful methodology to assess the metabolic phenotypes and capabilities of an organism. The starting point and cornerstone of all such modeling is a genome-scale metabolic network reconstruction. The creation, further development, and application of such networks is a growing field of research thanks to a plethora of readily accessible computational tools. While the majority of studies are focused on single-species analyses, typically of a microbe, the computational study of communities of organisms is gaining attention. Similarly, reconstructions that are unified for a multi-cellular organism have gained in popularity. Consequently, the rapid generation of genome-scale metabolic reconstructed networks is crucial. While multiple web-based or stand-alone tools are available for automated network reconstruction, there is, however, currently no publicly available tool that allows the swift assembly of draft reconstructions of community metabolic networks and consolidated metabolic networks for a specified list of organisms. RESULTS: Here, we present AutoKEGGRec, an automated tool that creates first draft metabolic network reconstructions of single organisms, community reconstructions based on a list of organisms, and finally a consolidated reconstruction for a list of organisms or strains. AutoKEGGRec is developed in Matlab and works seamlessly with the COBRA Toolbox v3, and it is based on only using the KEGG database as external input. The generated first draft reconstructions are stored in SBML files and consist of all reactions for a KEGG organism ID and corresponding linked genes. This provides a comprehensive starting point for further refinement and curation using the host of COBRA toolbox functions or other preferred tools. Through the data structures created, the tool also facilitates a comparative analysis of metabolic content in any given number of organisms present in the KEGG database. CONCLUSION: AutoKEGGRec provides a first step in a metabolic network reconstruction process, filling a gap for tools creating community and consolidated metabolic networks. Based only on KEGG data as external input, the generated reconstructions consist of data with a directly traceable foundation and pedigree. With AutoKEGGRec, this kind of modeling is made accessible to a wider part of the genome-scale metabolic analysis community.

wTO: an R package for computing weighted topological overlap and a consensus network with integrated visualization tool.

BACKGROUND: Network analyses, such as of gene co-expression networks, metabolic networks and ecological networks have become a central approach for the systems-level study of biological data. Several software packages exist for generating and analyzing such networks, either from correlation scores or the absolute value of a transformed score called weighted topological overlap (wTO). However, since gene regulatory processes can up- or down-regulate genes, it is of great interest to explicitly consider both positive and negative correlations when constructing a gene co-expression network. RESULTS: Here, we present an R package for calculating the weighted topological overlap (wTO), that, in contrast to existing packages, explicitly addresses the sign of the wTO values, and is thus especially valuable for the analysis of gene regulatory networks. The package includes the calculation of p-values (raw and adjusted) for each pairwise gene score. Our package also allows the calculation of networks from time series (without replicates). Since networks from independent datasets (biological repeats or related studies) are not the same due to technical and biological noise in the data, we additionally, incorporated a novel method for calculating a consensus network (CN) from two or more networks into our R package. To graphically inspect the resulting networks, the R package contains a visualization tool, which allows for the direct network manipulation and access of node and link information. When testing the package on a standard laptop computer, we can conduct all calculations for systems of more than 20,000 genes in under two hours. We compare our new wTO package to state of art packages and demonstrate the application of the wTO and CN functions using 3 independently derived datasets from healthy human pre-frontal cortex samples. To showcase an example for the time series application we utilized a metagenomics data set. CONCLUSION: In this work, we developed a software package that allows the computation of wTO networks, CNs and a visualization tool in the R statistical environment. It is publicly available on CRAN repositories under the GPL -2 Open Source License ( https://cran.r-project.org/web/packages/wTO/ ).

A composite network of conserved and tissue specific gene interactions reveals possible genetic interactions in glioma.

Differential co-expression network analyses have recently become an important step in the investigation of cellular differentiation and dysfunctional gene-regulation in cell and tissue disease-states. The resulting networks have been analyzed to identify and understand pathways associated with disorders, or to infer molecular interactions. However, existing methods for differential co-expression network analysis are unable to distinguish between various forms of differential co-expression. To close this gap, here we define the three different kinds (conserved, specific, and differentiated) of differential co-expression and present a systematic framework, CSD, for differential co-expression network analysis that incorporates these interactions on an equal footing. In addition, our method includes a subsampling strategy to estimate the variance of co-expressions. Our framework is applicable to a wide variety of cases, such as the study of differential co-expression networks between healthy and disease states, before and after treatments, or between species. Applying the CSD approach to a published gene-expression data set of cerebral cortex and basal ganglia samples from healthy individuals, we find that the resulting CSD network is enriched in genes associated with cognitive function, signaling pathways involving compounds with well-known roles in the central nervous system, as well as certain neurological diseases. From the CSD analysis, we identify a set of prominent hubs of differential co-expression, whose neighborhood contains a substantial number of genes associated with glioblastoma. The resulting gene-sets identified by our CSD analysis also contain many genes that so far have not been recognized as having a role in glioblastoma, but are good candidates for further studies. CSD may thus aid in hypothesis-generation for functional disease-associations.

Alginate Biosynthesis Factories in Pseudomonas fluorescens: Localization and Correlation with Alginate Production Level.

Pseudomonas fluorescens is able to produce the medically and industrially important exopolysaccharide alginate. The proteins involved in alginate biosynthesis and secretion form a multiprotein complex spanning the inner and outer membranes. In the present study, we developed a method by which the porin AlgE was detected by immunogold labeling and transmission electron microscopy. Localization of the AlgE protein was found to depend on the presence of other proteins in the multiprotein complex. No correlation was found between the number of alginate factories and the alginate production level, nor were the numbers of these factories affected in an algC mutant that is unable to produce the precursor needed for alginate biosynthesis. Precursor availability and growth phase thus seem to be the main determinants for the alginate production rate in our strain. Clustering analysis demonstrated that the alginate multiprotein complexes were not distributed randomly over the entire outer cell membrane surface.

Genome-wide association studies reveal differences in genetic susceptibility between single events vs. recurrent events of atrial fibrillation and myocardial infarction: the HUNT study.

Genetic research into atrial fibrillation (AF) and myocardial infarction (MI) has predominantly focused on comparing afflicted individuals with their healthy counterparts. However, this approach lacks granularity, thus overlooking subtleties within patient populations. In this study, we explore the distinction between AF and MI patients who experience only a single disease event and those experiencing recurrent events. Integrating hospital records, questionnaire data, clinical measurements, and genetic data from more than 500,000 HUNT and United Kingdom Biobank participants, we compare both clinical and genetic characteristics between the two groups using genome-wide association studies (GWAS) meta-analyses, phenome-wide association studies (PheWAS) analyses, and gene co-expression networks. We found that the two groups of patients differ in both clinical characteristics and genetic risks. More specifically, recurrent AF patients are significantly younger and have better baseline health, in terms of reduced cholesterol and blood pressure, than single AF patients. Also, the results of the GWAS meta-analysis indicate that recurrent AF patients seem to be at greater genetic risk for recurrent events. The PheWAS and gene co-expression network analyses highlight differences in the functions associated with the sets of single nucleotide polymorphisms (SNPs) and genes for the two groups. However, for MI patients, we found that those experiencing single events are significantly younger and have better baseline health than those with recurrent MI, yet they exhibit higher genetic risk. The GWAS meta-analysis mostly identifies genetic regions uniquely associated with single MI, and the PheWAS analysis and gene co-expression networks support the genetic differences between the single MI and recurrent MI groups. In conclusion, this work has identified novel genetic regions uniquely associated with single MI and related PheWAS analyses, as well as gene co-expression networks that support the genetic differences between the patient subgroups of single and recurrent occurrence for both MI and AF.

Parameter inference for enzyme and temperature constrained genome-scale models.

The metabolism of all living organisms is dependent on temperature, and therefore, having a good method to predict temperature effects at a system level is of importance. A recently developed Bayesian computational framework for enzyme and temperature constrained genome-scale models (etcGEM) predicts the temperature dependence of an organism's metabolic network from thermodynamic properties of the metabolic enzymes, markedly expanding the scope and applicability of constraint-based metabolic modelling. Here, we show that the Bayesian calculation method for inferring parameters for an etcGEM is unstable and unable to estimate the posterior distribution. The Bayesian calculation method assumes that the posterior distribution is unimodal, and thus fails due to the multimodality of the problem. To remedy this problem, we developed an evolutionary algorithm which is able to obtain a diversity of solutions in this multimodal parameter space. We quantified the phenotypic consequences on six metabolic network signature reactions of the different parameter solutions resulting from use of the evolutionary algorithm. While two of these reactions showed little phenotypic variation between the solutions, the remainder displayed huge variation in flux-carrying capacity. This result indicates that the model is under-determined given current experimental data and that more data is required to narrow down the model predictions. Finally, we made improvements to the software to reduce the running time of the parameter set evaluations by a factor of 8.5, allowing for obtaining results faster and with less computational resources.