SUMMARY: Accurate enzyme turnover numbers are essential for building enzyme-constrained genome-scale metabolic models. However, collecting and curating these parameters remains a major bottleneck. Indeed, kcat values are scattered across multiple databases, reported under varying experimental conditions, and often missing for many enzymes. To address this challenge, we present WILDkCAT, a Python-based pipeline that enables the retrieval of kcat values from wild-type enzyme measured under user-specified pH and temperature ranges for a given metabolic model. The application to Escherichia coli (iML1515) and Homo sapiens (Human-GEM) models demonstrated the ability of WILDkCAT to retrieve substantial kcat coverage and its applicability across diverse genome-scale models. AVAILABILITY AND IMPLEMENTATION: WILDkCAT is available at https://github.com/sysbiolux/WILDkCAT and from PyPI. WILDkCAT works on all major operating systems and computer architectures. The documentation is available at https://sysbiolux.github.io/WILDkCAT.

Gliomas are the most common type of malignant brain tumors, with glioblastoma multiforme (GBM) having a median survival of 15 months due to drug resistance and relapse. The treatment of gliomas relies on surgery, radiotherapy and chemotherapy. Only 12 anti-brain tumor chemotherapies (AntiBCs), mostly alkylating agents, have been approved so far. Glioma subtype-specific metabolic models were reconstructed to simulate metabolite exchanges, in silico knockouts and the prediction of drug and drug combinations for all three subtypes. The simulations were confronted with literature, high-throughput screenings (HTSs), xenograft and clinical trial data to validate the workflow and further prioritize the drug candidates. The three subtype models accurately displayed different degrees of dependencies toward glutamine and glutamate. Furthermore, 33 single drugs, mainly antimetabolites and TXNRD1-inhibitors, as well as 17 drug combinations were predicted as potential candidates for gliomas. Half of these drug candidates have been previously tested in HTSs. Half of the tested drug candidates reduce proliferation in cell lines and two-thirds in xenografts. Most combinations were predicted to be efficient for all three glioma types. However, eflornithine/rifamycin and cannabidiol/adapalene were predicted specifically for GBM and low-grade glioma, respectively. Most drug candidates had comparable efficiency in preclinical tests, cerebrospinal fluid bioavailability and mode-of-action to AntiBCs. However, fotemustine and valganciclovir alone and eflornithine and celecoxib in combination with AntiBCs improved the survival compared to AntiBCs in two-arms, phase I/II and higher glioma clinical trials. Our work highlights the potential of metabolic modeling in advancing glioma drug discovery, which accurately predicted metabolic vulnerabilities, repurposable drugs and combinations for the glioma subtypes.

Stratification of patients diagnosed with cancer has become a major goal in personalized oncology. One important aspect is the accurate prediction of the response to various drugs. It is expected that the molecular characteristics of the cancer cells contain enough information to retrieve specific signatures, allowing for accurate predictions based solely on these multi-omic data. Ideally, these predictions should be explainable to clinicians, in order to be integrated in the patients care. We propose a machine-learning framework based on ensemble learning to integrate multi-omic data and predict sensitivity to an array of commonly used and experimental compounds, including chemotoxic compounds and targeted kinase inhibitors. We trained a set of classifiers on the different parts of our dataset to produce omic-specific signatures, then trained a random forest classifier on these signatures to predict drug responsiveness. We used the Cancer Cell Line Encyclopedia dataset, comprising multi-omic and drug sensitivity measurements for hundreds of cell lines, to build the predictive models, and validated the results using nested cross-validation. Our results show good performance for several compounds (Area under the Receiver-Operating Curve >79%) across the most frequent cancer types. Furthermore, the simplicity of our approach allows to examine which omic layers have a greater importance in the models and identify new putative markers of drug responsiveness. We propose several models based on small subsets of transcriptional markers with the potential to become useful tools in personalized oncology, paving the way for clinicians to use the molecular characteristics of the tumors to predict sensitivity to therapeutic compounds.

Project-based learning (PBL) is a dynamic student-centred teaching method that encourages students to solve real-life problems while fostering engagement and critical thinking. Here, we report on a PBL course on metabolic network modelling that has been running for several years within the Master in Integrated Systems Biology (MISB) at the University of Luxembourg. This 2-week full-time block course comprises an introduction into the core concepts and methods of constraint-based modelling (CBM), applied to toy models and large-scale networks alongside the preparation of individual student projects in week 1 and, in week 2, the presentation and execution of these projects. We describe in detail the schedule and content of the course, exemplary student projects, and reflect on outcomes and lessons learned. PBL requires the full engagement of students and teachers and gives a rewarding teaching experience. The presented course can serve as a role model and inspiration for other similar courses.

Drug repositioning has received increased attention since the past decade as several blockbuster drugs have come out of repositioning. Computational approaches are significantly contributing to these efforts, of which, network-based methods play a key role. Various structural (topological) network measures have thereby contributed to uncovering unintuitive functional relationships and repositioning candidates in drug-disease and other networks. This review gives a broad overview of the topic, and offers perspectives on the application of topological measures for network analysis. It also discusses unexplored measures, and draws attention to a wider scope of application efforts, especially in drug repositioning.

MOTIVATION: Mathematical modelling of regulatory networks allows for the discovery of knowledge at the system level. However, existing modelling tools are often computation-heavy and do not offer intuitive ways to explore the model, to test hypotheses or to interpret the results biologically. RESULTS: We have developed a computational approach to contextualize logical models of regulatory networks with biological measurements based on a probabilistic description of rule-based interactions between the different molecules. Here, we propose a Matlab toolbox, FALCON, to automatically and efficiently build and contextualize networks, which includes a pipeline for conducting parameter analysis, knockouts and easy and fast model investigation. The contextualized models could then provide qualitative and quantitative information about the network and suggest hypotheses about biological processes. AVAILABILITY AND IMPLEMENTATION: FALCON is freely available for non-commercial users on GitHub under the GPLv3 licence. The toolbox, installation instructions, full documentation and test datasets are available at https://github.com/sysbiolux/FALCON. FALCON runs under Matlab (MathWorks) and requires the Optimization Toolbox. CONTACT: [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Genome-scale metabolic network reconstructions provide a basis for the investigation of the metabolic properties of an organism. There are reconstructions available for multiple organisms, from prokaryotes to higher organisms and methods for the analysis of a reconstruction. One example is the use of flux balance analysis to improve the yields of a target chemical, which has been applied successfully. However, comparison of results between existing reconstructions and models presents a challenge because of the heterogeneity of the available reconstructions, for example, of standards for presenting gene-protein-reaction associations, nomenclature of metabolites and reactions or selection of protonation states. The lack of comparability for gene identifiers or model-specific reactions without annotated evidence often leads to the creation of a new model from scratch, as data cannot be properly matched otherwise. In this contribution, we propose to improve the predictive power of metabolic models by switching from gene-protein-reaction associations to transcript-isoform-reaction associations, thus taking advantage of the improvement of precision in gene expression measurements. To achieve this precision, we discuss available databases that can be used to retrieve this type of information and point at issues that can arise from their neglect. Further, we stress issues that arise from non-standardized building pipelines, like inconsistencies in protonation states. In addition, problems arising from the use of non-specific cofactors, e.g. artificial futile cycles, are discussed, and finally efforts of the metabolic modelling community to unify model reconstructions are highlighted.

The transcription factor nuclear factor kappa-B (NFκB) is a key regulator of pro-inflammatory and pro-proliferative processes. Accordingly, uncontrolled NFκB activity may contribute to the development of severe diseases when the regulatory system is impaired. Since NFκB can be triggered by a huge variety of inflammatory, pro-and anti-apoptotic stimuli, its activation underlies a complex and tightly regulated signaling network that also includes multi-layered negative feedback mechanisms. Detailed understanding of this complex signaling network is mandatory to identify sensitive parameters that may serve as targets for therapeutic interventions. While many details about canonical and non-canonical NFκB activation have been investigated, less is known about cellular IκBα pools that may tune the cellular NFκB levels. IκBα has so far exclusively been described to exist in two different forms within the cell: stably bound to NFκB or, very transiently, as unbound protein. We created a detailed mathematical model to quantitatively capture and analyze the time-resolved network behavior. By iterative refinement with numerous biological experiments, we yielded a highly identifiable model with superior predictive power which led to the hypothesis of an NFκB-lacking IκBα complex that contains stabilizing IKK subunits. We provide evidence that other but canonical pathways exist that may affect the cellular IκBα status. This additional IκBα:IKKγ complex revealed may serve as storage for the inhibitor to antagonize undesired NFκB activation under physiological and pathophysiological conditions.

Systemic approaches to the study of a biological cell or tissue rely increasingly on the use of context-specific metabolic network models. The reconstruction of such a model from high-throughput data can routinely involve large numbers of tests under different conditions and extensive parameter tuning, which calls for fast algorithms. We present fastcore, a generic algorithm for reconstructing context-specific metabolic network models from global genome-wide metabolic network models such as Recon X. fastcore takes as input a core set of reactions that are known to be active in the context of interest (e.g., cell or tissue), and it searches for a flux consistent subnetwork of the global network that contains all reactions from the core set and a minimal set of additional reactions. Our key observation is that a minimal consistent reconstruction can be defined via a set of sparse modes of the global network, and fastcore iteratively computes such a set via a series of linear programs. Experiments on liver data demonstrate speedups of several orders of magnitude, and significantly more compact reconstructions, over a rival method. Given its simplicity and its excellent performance, fastcore can form the backbone of many future metabolic network reconstruction algorithms.

Apoptosis is regulated by several signaling pathways which are extensively linked by crosstalks. Boolean or logical modeling has become a promising approach to capture the qualitative behavior of such complex networks. Here we built a large-scale literature-based Boolean model of the central intrinsic and extrinsic apoptosis pathways as well as pathways connected with them. The model responds to several external stimuli such as Fas ligand, TNF-alpha, UV-B irradiation, interleukin-1beta and insulin. Timescales and multi-value node logic were used and turned out to be indispensable to reproduce the behavior of the apoptotic network. The coherence of the model was experimentally validated. Thereby an UV-B dose-effect is shown for the first time in mouse hepatocytes. Analysis of the model revealed a tight regulation emerging from high connectivity and spanning crosstalks and a particular importance of feedback loops. An unexpected feedback from Smac release to RIP could further increase complex II formation. The introduced Boolean model provides a comprehensive and coherent description of the apoptosis network behavior. It gives new insights into the complex interplay of pro- and antiapoptotic factors and can be easily expanded to other signaling pathways.

BACKGROUND: Receptors and scaffold proteins possess a number of distinct domains and bind multiple partners. A common problem in modeling signaling systems arises from a combinatorial explosion of different states generated by feasible molecular species. The number of possible species grows exponentially with the number of different docking sites and can easily reach several millions. Models accounting for this combinatorial variety become impractical for many applications. RESULTS: Our results show that under realistic assumptions on domain interactions, the dynamics of signaling pathways can be exactly described by reduced, hierarchically structured models. The method presented here provides a rigorous way to model a large class of signaling networks using macro-states (macroscopic quantities such as the levels of occupancy of the binding domains) instead of micro-states (concentrations of individual species). The method is described using generic multidomain proteins and is applied to the molecule LAT. CONCLUSION: The presented method is a systematic and powerful tool to derive reduced model structures describing the dynamics of multiprotein complex formation accurately.