Search

Genome-scale, constraint-based models (GEM) and their derivatives are commonly used to model and gain insights into microbial metabolism. Often, however, their accuracy and predictive power are limited and enable only approximate designs. To improve their usefulness for strain and bio-process design, we studied here their capacity to accurately predict metabolic changes in response to operational conditions in a bioreactor, as well as intracellular, active reactions. We used flux balance analysis (FBA) and dynamic FBA (dFBA) to predict growth dynamics of the model organism Saccharomyces cerevisiae under different industrially relevant conditions. We compared simulations with the latest developed GEM for this organism (Yeast8) and its enzyme-constrained version (ecYeast8) herein described with experimental data and found that ecYeast8 outperforms Yeast8 in all the simulations. EcYeast8 was able to predict well-known traits of yeast metabolism including the onset of the Crabtree effect, the order of substrate consumption during mixed carbon cultivation and production of a target metabolite. We showed how the combination of ecGEM and dFBA links reactor operation and genetic modifications to flux predictions, enabling the prediction of yields and productivities of different strains and (dynamic) production processes. Additionally, we present flux sampling as a tool to analyze flux predictions of ecGEM, of major importance for strain design applications. We showed that constraining protein availability substantially improves accuracy of the description of the metabolic state of the cell under dynamic conditions. This therefore enables more realistic and faithful designs of industrially relevant cell-based processes and, thus, the usefulness of such models

Atlantic salmon (Salmo salar) is the most valuable farmed fish globally and there is much interest in optimizing its genetics and rearing conditions for growth and feed efficiency. Marine feed ingredients must be replaced to meet global demand, with challenges for fish health and sustainability. Metabolic models can address this by connecting genomes to metabolism, which converts nutrients in the feed to energy and biomass, but such models are currently not available for major aquaculture species such as salmon. We present SALARECON, a model focusing on energy, amino acid, and nucleotide metabolism that links the Atlantic salmon genome to metabolic fluxes and growth. It performs well in standardized tests and captures expected metabolic (in)capabilities. We show that it can explain observed hypoxic growth in terms of metabolic fluxes and apply it to aquaculture by simulating growth with commercial feed ingredients. Predicted limiting amino acids and feed efficiencies agree with data, and the model suggests that marine feed efficiency can be achieved by supplementing a few amino acids to plant- and insect-based feeds. SALARECON is a high-quality model that makes it possible to simulate Atlantic salmon metabolism and growth. It can be used to explain Atlantic salmon physiology and address key challenges in aquaculture such as development of sustainable feeds.Author summaryAtlantic salmon aquaculture generates billions of euros annually, but faces challenges of sustainability. Salmon are carnivores by nature, and fish oil and fish meal have become scarce resources in fish feed production. Novel, sustainable feedstuffs are being trialed hand in hand with studies of the genetics of growth and feed efficiency. This calls for a mathematical-biological framework to integrate data with understanding of the effects of novel feeds on salmon physiology and its interplay with genetics. We have developed the SALARECON model of the core salmon metabolic reaction network, linking its genome to metabolic fluxes and growth. Computational analyses show good agreement with observed growth, amino acid limitations, and feed efficiencies, illustrating the potential for in silico studies of potential feed mixtures. In particular, in silico screening of possible diets will enable more efficient animal experiments with improved knowledge gain. We have adopted best practices for test-driven development, virtual experiments to assay metabolic capabilities, revision control, and FAIR data and model management. This facilitates fast, collaborative, reliable development of the model for future applications in sustainable production biology.

Constraint-based, genome-scale metabolic models are an essential tool to guide metabolic engineering. However, they lack the detail and time dimension that kinetic models with enzyme dynamics offer. Model reduction can be used to bridge the gap between the two methods and allow for the integration of kinetic models into the Design-Built-Test-Learn cycle. Here we show that these reduced size models can be representative of the dynamics of the original model and demonstrate the automated generation and parameterisation of such models. Using these minimal models of metabolism could allow for further exploration of dynamic responses in metabolic networks., Metabolic Engineering, 64, ISSN:1096-7176, ISSN:1096-7184

The microalgae Tetradesmus obliquus is able to maintain a high photosynthetic efficiency under nitrogen limitation and is considered a promising green microalgae for sustainable production of diverse compounds, including biofuels. Here, we report the first draft whole-genome shotgun sequencing of T. obliquus . The final assembly comprises 108,715,903 bp with over 1,368 scaffolds.

Background The development of biological processes that replace the existing petrochemical-based industry is one of the biggest challenges in biotechnology. Aspergillus niger is one of the main industrial producers of lignocellulolytic enzymes, which are used in the conversion of lignocellulosic feedstocks into fermentable sugars. Both the hydrolytic enzymes responsible for lignocellulose depolymerisation and the molecular mechanisms controlling their expression have been well described, but little is known about the transport systems for sugar uptake in A. niger. Understanding the transportome of A. niger is essential to achieve further improvements at strain and process design level. Therefore, this study aims to identify and classify A. niger sugar transporters, using newly developed tools for in silico and in vivo analysis of its membrane-associated proteome. Results In the present research work, a hidden Markov model (HMM), that shows a good performance in the identification and segmentation of functionally validated glucose transporters, was constructed. The model (HMMgluT) was used to analyse the A. niger membrane-associated proteome response to high and low glucose concentrations at a low pH. By combining the abundance patterns of the proteins found in the A. niger plasmalemma proteome with their HMMgluT scores, two new putative high-affinity glucose transporters, denoted MstG and MstH, were identified. MstG and MstH were functionally validated and biochemically characterised by heterologous expression in a S. cerevisiae glucose transport null mutant. They were shown to be a high-affinity glucose transporter (Km = 0.5 ± 0.04 mM) and a very high-affinity glucose transporter (Km = 0.06 ± 0.005 mM), respectively. Conclusions This study, focusing for the first time on the membrane-associated proteome of the industrially relevant organism A. niger, shows the global response of the transportome to the availability of different glucose concentrations. Analysis of the A. niger transportome with the newly developed HMMgluT showed to be an efficient approach for the identification and classification of new glucose transporters. Electronic supplementary material The online version of this article (doi:10.1186/s13068-015-0317-9) contains supplementary material, which is available to authorized users.

Systems biology requires the availability, co-ordination and simultaneous interaction of a large number of diverse facilities and activities. These cover an entire spectrum, from mathematical modelling, through biological, biomedical and clinical experiments, to dedicated technology development. The systems biology community needs close cooperation with data-generation groups and bioinformaticians to define a strategy for producing life-science data of sufficiently high quality for model generation. For each medical, biological or biotechnological problem addressed, the optimal combination of facilities and activities is likely to be different. The complexity of biological systems, and the diversity and dynamics of their processes, means that a full analysis is far too complex to be handled by a single entity, industry or country – a variety of specialist expertise and facilities are typically necessary to achieve results suitable for modelling. Systems-level approaches for tackling the complexity of life-science data provide a profound conceptual advance compared to reductionist biological research methods of the past. Rather than focusing on individual laboratories, specialising in a limited number of research technologies, the Infrastructure for Systems Biology in Europe (ISBE) will facilitate the synergistic application of a wide range of research techniques and technologies to problems of major medical and biotechnological importance.

A limited life cycle assessment (LCA) was performed on a combined biological and chemical process for the production of adipic acid, which was compared to the traditional petrochemical process. The LCA comprises the biological conversion of the aromatic feedstocks benzoic acid, impure aromatics, toluene, or phenol from lignin to cis, cis-muconic acid, which is subsequently converted to adipic acid through hydrogenation. Apart from the impact of usage of petrochemical and biomass-based feedstocks, the environmental impact of the final concentration of cis, cis-muconic acid in the fermentation broth was studied using 1.85% and 4.26% cis, cis-muconic acid. The LCA focused on the cumulative energy demand (CED), cumulative exergy demand (CExD), and the CO(2) equivalent (CO(2) eq) emission, with CO(2) and N(2) O measured separately. The highest calculated reduction potential of CED and CExD were achieved using phenol, which reduced the CED by 29% and 57% with 1.85% and 4.26% cis, cis-muconic acid, respectively. A decrease in the CO(2) eq emission was especially achieved when the N(2) O emission in the combined biological and chemical process was restricted. At 4.26% cis, cis-muconic acid, the different carbon backbone feedstocks contributed to an optimized reduction of CO(2) eq emissions ranging from 14.0 to 17.4 ton CO(2) eq/ton adipic acid. The bulk of the bioprocessing energy intensity is attributed to the hydrogenation reactor, which has a high environmental impact and a direct relationship with the product concentration in the broth.

Pseudomonas putida KT2440-JD1 was derived from P. putida KT2440 after N-methyl-N'-nitro-N-nitrosoguanidine (NTG)-mutagenesis and exposure to 3-fluorobenzoate (3-FB). The mutant was no longer able to grow using benzoate as a sole carbon source, but co-metabolized benzoate to cis, cis-muconate during growth on glucose, which accumulated in the growth medium. The specific production rate (qpm) was 0.18±0.03g cis, cis-muconate/(gDCWh) in continuous cultures, and increased to 1.4g cis, cis-muconate/(gDCWh) during wash-out cultivation. Transcriptome analysis showed that the cat operon was not induced in P. putida KT2440-JD1 in the presence of 5mM benzoate, due to a point mutation in the highly conserved DNA binding domain of the transcriptional regulator (catR) of the cat operon. The ben operon was highly expressed in the presence of benzoate in the mutant and its parental strain. This operon contains PP_3166 (catA2), which was shown to be a second catechol 1,2-dioxygenase besides catA. P. putida KT2440-JD1 is the first cis, cis-muconate-accumulating mutant that was characterized at the genetic level. The specific production rate achieved is at least eight times higher than those reported for other cis, cis-muconate-producing strains. © 2011 Elsevier B.V.

A novel modeling approach for the description of bioprocesses is proposed, linking microbial growth kinetics to gene regulation. An example is given with the development and experimental validation of a dynamic mathematical model of the TOL plasmid of Pseudomonas putida mt-2, which is used for the metabolism of m-xylene. The model of this genetic circuit is coupled to a growth kinetic model through predictions of rate-limiting enzyme concentrations that control biomass growth and substrate consumption. Batch cultures of mt-2 fed with m-xylene were performed to estimate model parameters and to confirm that the combined model successfully describes the bioprocess, through mRNA, biomass and m-xylene concentration measurements. However, mathematical models developed exclusively based on macroscopic measurements failed to predict the process variables, highlighting the importance of gene regulation for the development of advanced biological models. © 2011 Elsevier B.V.

A modelling framework that consists of model building, validation and analysis, leading to model-based design of experiments and to the application of optimisation-based model-predictive control strategies for the development of optimised bioprocesses is presented. An example of this framework is given with the construction and experimental validation of a dynamic mathematical model of the Ps/Pr promoters system of the TOL plasmid, which is used for the metabolism of m -xylene by Pseudomonas putida mt-2. Furthermore, the genetic circuit model is combined with the growth kinetics of the strain in batch cultures, demonstrating how the description of key genetic circuits can facilitate the improvement of existing growth kinetic models that fail to predict unusual growth patterns. Consequently, the dynamic model is combined with global sensitivity analysis, which is used to identify the presence of significant model parameters, constituting a model-based methodology for the formulation of genetic circuit optimization methods.

In his editorial overview for the 2008 Special Issue on this topic, the late Jaroslav Stark pointedly noted that systems biology is no longer a niche pursuit, but a recognized discipline in its own right “noisily” coming of age [1]. Whilst general underlying principles and basic techniques are now reasonably well established, we see a progressive shift from “fishing expeditions” through seas of “omics” data towards increasingly more model-driven research generating verifiable hypotheses that are iteratively tested and refined across different aggregation levels. In this special issue, we illustrate this ethos of systems biology at work in various fields through an eclectic series of articles thematically grouped under “methods and tools in network analysis”, “emerging and enabling technologies”, “microbes in disease and the environment”, “human health and disease” and “synthetic biology”.

This paper gives an overview of the most recent advances in the development of compact systems for integrated nitrogen removal. First, a brief glance at the conventional configurations most commonly used is given. Compact systems based on immobilized-cell processes will then be discussed on the basis of their performance, adequacy and potential applicability on a large scale. Novel nitrification-denitrification processes will be evaluated as well as some important tools (such as modeling and its experimental evaluation by several ways) for the study of nitrogen removal in compact reactors. Finally, the future prospects on integrated nitrogen removal will be discussed. The comparative analysis made of many advanced systems indicates that those that are the most compact and effective involve immobilized-cell technology (either naturally attached or artificially immobilized) and tend to be designed in height. This is just the opposite of the treatment plants currently in operation that are based chiefly in activated-sludge systems (suspended bacteria in large shallow tanks). Another important tendency is the attempt to integrate nitrification and denitrifica-tion within the most compact systems possible. The extreme cases are the systems in which both steps occur within the same biocatalyst particle. One of the most promising configurations found was that in which nitrogen removal proceeds fully autotrophically within a single carrier.

Synthetic biologists design and engineer organisms for a better and more sustainable future based on a currently transitioning global economy grounded in biomanufacturing. While the prospects are encouraging, concerns about uncertainty and risks associated to genome editing, including uncontrolled proliferation of transgenic microorganisms, need to be considered in safe-by-design bioengineering strategies. This dissertation delves into different types of genetic safeguards to conditionally restrict cell host viability to defined environments using the popular metabolic engineering hostPseudomonas putidaKT2440. Specifically, this thesis encompasses the generation of an industrially appealing synthetic phosphite auxotroph attained via metabolic rewiring, and a CRISPR-Cas9-based kill switch based on different genetic circuits that react to external signals for the activation of a highly genotoxic response. To also prevent horizontal gene transfer, we developed ReScribe, a highly optimized recombineering tool enhanced by CRISPR-ScCas9-mediated counterselection that was used to generate a minimally recodedP. putidastrain of essential metabolic genes. To further improve our genome editing capabilities, we employed a serial enrichment workflow followed by a next-generation sequencing analysis to screen in a high-throughput manner a library of potential recombineering systems in differentPseudomonasspecies, eventually finding new variants for enhanced mutagenesis. While all these genetic safeguards and tools are attractive in theory, their implementation in industrial biotechnology and real-world applications is rarely exercised in practice, despite technological advances. In this dissertation we question why and what could be done about it and claim that an explicit strategy of contextualization, that is, an early emphasis on potential applications, can assist the development of genetic safeguards. Moreover, the conduction of a series of interviews and analysis inspired by social science methodologies resulted in different explorations of research and ethical questions. These included the learning of safety and safe-by-design approaches in the life sciences programmes at Wageningen University, the impact of tensions in stakeholder norms on designing for biosafety in the industrial biotechnology context, and the perceptions of industry and academia on production platforms and opportunities in the biotechnology field. Lastly, this thesis ends providing a snapshot of the state of biomanufacturing and proposing the idea of meta-workflows. These are unifying ecosystems amongst research infrastructures that emphasize the importance of interoperability, harmonization, and sustainability in making a successful transition to global manufacturing that utilizes bio-based and safe production platforms. &nbsp

Exceeding planetary boundaries, and in particular climate change, is quickly forcing the global economy to align with the principles of sustainability. In this context, there is a strong call from policy makers, scientists, and citizens to step out of the oil-based era and relentlessly pursue a biobased, sustainable economy. The industrial exploitation of microbes for the sustainable production of bio-based chemicals and fuels has contributed significantly to this effort. Nevertheless, it is becoming clear that industrial biotechnology requires a wider repertoire of more effective microbial hosts to expand and enhance its capabilities. At the same time, advancements in synthetic biology allow the rapid and efficient domestication of several unconventional microbes with appealing industrial characteristics. One bacterial host that has gained increasing attention as an efficient microbial cell factory is Pseudomonas putida KT2440. However, despite its many attractive inherent features, additional efforts must be made to elucidate and engineer its complex metabolic and regulatory architecture. Hence, this thesis aimed at contributing to the industrial domestication of P. putida by developing and applying cutting-edge synthetic biology technologies for engineering its cellular metabolism.The first part of the research herein was focused on the development of genetic tools, namely: i) the SEVA 3.1 vector library, optimized for Gram-negative bacteria, built upon the principles of the SEVA, BioBricks, and Type IIS restriction enzyme standards; and ii) the first tunable CRISPRi system for gene repression in P. putida. The second part aimed at constructing three P. putida strains tailored to overproduce the key precursor metabolites pyruvate, malonyl-CoA, and chorismate. Each strain was approached from a different metabolic engineering perspective dealing with the concepts of dynamic growth control, high-throughput optimization, and directed adaptive laboratory evolution. The work performed follows the principles of the DBTL cycle using in combination innovative computational and genetic methods that can be readily applied to further expand the industrial potential of P. putida.

Proteins are key to many different processes in and around living cells, they are the molecular mechanisms that facilitate life. Therefore, protein function and structure have been of great interest to researchers over the past decades. A better understanding of how proteins work is important, as it will help us understand the various diseases caused by their malfunction and help us apply proteins in various applications, such as industrial processes. The research on proteins has resulted in an ever-increasing amount of data related to proteins.Protein information systems were developed to make sense out of all the data available for specific protein classes. The work described in this thesis makes use of the 3DM platform to store and order the enormous amount of protein data that is nowadays available. The 3DM platform integrates heterogeneous datasets for complete protein superfamilies to generate molecular class specific information systems (3DM systems).The main objective of this thesis is to facilitate a better understanding of protein function through data integration- and analysis-tools to support both DNA diagnostics and protein engineering efforts. To that end, chapter 2 and chapter 3 focus on the collection and integration of mutational data from scientific literature, followed by protein superfamily analysis to reveal predictive patterns. In chapter 4, mutational data mined for whole families are combined with a superfamily alignment analysis to assign function to networks of residues that appear to have co-evolved. And in chapter 5, the data of a 3DM system is used to train a machine learning model for the prediction of pathogenicity of novel genetic variants.In chapter 2 a method for the automated collection and integration of mutation data from scientific literature is described. This text-mining tool, Mutator, was designed to automatically identify and retrieve relevant literature for a complete protein family.Mutator was tested on the alpha-amylase protein family which contains a protein that, when mutated, can lead to Fabry’s disease. The Fabry disease mutation database was constructed which contained 30% more mutation data when compared to the manually curated HGMD. To differentiate between natural variation and pathogenic mutations, alignment patterns in the alpha-amylase superfamily were investigated and Validator was developed, a tool to evaluate the potential pathogenicity of individual mutations.In a follow up study, chapter 3, nine disease-related proteins across two different protein families are investigated. An improved version of Mutator was developed and used to collect mutational data for these protein families. A comparison to the manually curated HGMD showed that Mutator extracts far more mutation data, both in terms of unique mutations extracted, as well as number of references per mutation.These mutation datasets were analyzed in the context of the protein families to identify patterns to assess pathogenicity for novel variants. This analysis showed that pathogenic mutations would more often be observed in the structurally conserved core of the family than in structurally variable regions. A significant overlap was observed of protein positions with pathogenic mutations between very distant homologs, indicating that data can be transferred between equivalent residue positions even across large phylogenetic distances. Protein superfamily information systems that use structural alignments, such as 3DM, can facilitate the transfer of data across equivalent residues of structurally conserved positions.Chapter 4 describes a web-based tool for the analysis of co-evolving residues in protein super-family alignments. This tool, CorNet, combines networks of co-evolving positions with literature data to shed light on the function of these residues. When keywords, such as “specificity”, are mentioned in the sentences that contain mutational data for an alignment position, these were used to annotate the alignment position of a network.For six different protein families, a list of literature keywords was used to annotate network positions and to calculate an enrichment score of any keyword to be specifically associated with a network of co-evolving positions. Enriched keywords indicate the function behind evolutionary pressures that caused networks of residues to co-evolve. It was shown that an assigned function can be transferred from annotated to non-annotated positions in a network with a mutagenesis study that targeted only the non-annotated positions of a network enriched for the keyword “enantioselectivity”.In chapter 5, the integrated data of 3DM systems is utilized to train a model to predict variant pathogenicity with machine learning techniques. For three genes related to Long QT syndrome and Brugada syndrome, both cardiac arrhythmia syndromes, datasets of both benign and pathogenic variants were collected. 3DM was used to generate much heterogeneous data to describe each variant. Using this data, a pathogenicity predictor was generated to separate disease-causing from non-disease causing variants. This predictor outperformed frequently used genome-wide prediction tools and even gene-family specific predictors.In the general discussion an outlook is given on the scale-up of protein information systems to cover the complete structural space and even the whole human exome. This thesis provides a variety of avenues to apply integrated data to gain insight in proteins and support protein related research and development.

Metabolic engineering is a cornerstone of the bio-based economy, which aims to replace fossil-fuel based processes with biological systems. In this thesis, I develop and apply computational methods for model-driven engineering of microbial metabolism. Due to the complexity of biological systems, it is often unclear beforehand exactly which aspects of the systems will be limiting when engineering metabolism. By using the data coming from experiments to inform metabolic models, these models can be used to drive decision-making for the subsequent set of experiments, thus solving obstacles one at a time with the most information possible; a process known as the Design-Build-Test-Learn cycle. Models are thus an important part of metabolic engineering and in this thesis I use several modelling frameworks: constraint-based optimization, flux sampling, as well as kinetic modelling. Each modelling method has its strengths and in chapter 2 I review the use of models in metabolic engineering, consolidating the lessons I learned by applying different modelling methods and working together with experimental partners. We discuss the various questions that should be posed before deciding on a modelling strategy, as well as the strengths and weaknesses of the most common frameworks used for metabolic modelling. In addition, we discuss the different sources of experimental data and their suitability for using them in the different modelling frameworks.In chapter 3, we apply constraint-based modelling to Pseudomonas putida to support the development of a metabolic valve enabling control over the growth versus production trade-off. Using an existing constraint-based genome-scale model for P. putida, we evaluate the potential of the PDH reaction, producing acetyl-CoA from pyruvate, to serve as this metabolic valve. However, the current model fails to recognize the experimentally confirmed essentiality of this reaction. Thus, we search for alternative pathways in the model that can cause this misprediction and show that with these reactions disabled, the PDH reaction can function as a metabolic valve, allowing growth or the production of pyruvate derived products depending on its activity. This prediction was experimentally verified with the production of pyruvate as well as two pyruvate-derived compounds: 2-ketoisovalerate and lycopene. In chapter 4, we continue our work in P. putida. Using flux sampling, we identify targets for genetic up- and down-regulation for the overproduction of malonyl-CoA, a precursor for the synthesis of fatty acids as well as polyketides. By comparing the simulated fluxes in growing versus producing cells, we find targets that undergo a significant change of flux. These predictions were experimentally verified using a biosensor-based high-throughput screening, as well as through the production of phloroglucinol, a compound derived from malonyl-CoA.With chapter 5 we turn our attention towards kinetic models. While kinetic models are promising, they are significantly harder to work with due to their increased demands for data, as well as difficulties in simulating and optimizing larger models. Thus, we develop a pipeline for the automated generation of these types of models and explore how well this automated approach holds up with missing or inaccurate estimates of the kinetic parameters. Finally, we investigate the use of automated model-reduction methods, which we further explore in chapter 6. Here, we focus on the topic of automated model reduction, as the scale at which constraint-based models can be applied is much larger than the applicable scale for kinetic models. We investigate how well several model reduction methods, operating on the constraint-based model, reproduce the behaviour of the full-sized kinetic model after estimating the parameters with a time-series data set. We find that these reduced models can be representative of the full-scale model and can thus be a useful proxy for engineering the whole system.In chapter 7, we return to P. putida for the production of curcumin, but now with the aid of a kinetic model. Curcumin is a yellow-coloured compound, originally isolated from the turmeric plant, used as a food additive and colouring agent. Furthermore, the precursors for the production of curcuminoids can be derived from lignin, a substantial component of plant waste biomass, making it an interesting target for the bio-based, circular economy. The curcumin production pathway utilizes only a few enzymes but can lead to several products depending on enzyme kinetics and relative concentrations of intermediates. This makes it an interesting test case for a kinetic model, as constraint-based models do not model these details due to their steady-state assumption. Through sensitivity analysis on the kinetic model, we perform optimal experimental design to predict the experiments most informative for improving the model used to study the pathway.Finally, in chapter 8 I discuss how these different chapters contributed to the objectives of this thesis, and look towards how to further integrate computational models of metabolism into metabolic engineering studies in the future. In addition, I elaborate on challenges and future opportunities for using kinetic metabolic models.

The study of genetically identical cells frequently reveals that substantial variation exists between the expression levels of molecules, such as mRNAs and proteins, with respect to time in individual cells and across a population. This is due to intrinsic noise arising from the random timing of biochemical reactions in the gene regulatory network. An additional source of noise in biological systems comes from interactions of unknown molecular components with the network, which are collectively termed extrinsic noise. Mathematical modelling has been used to understand the factors that contribute to the resulting stochasticity in the gene expression dynamics of cells. As many molecular species are present in small numbers, we require a stochastic description of these biochemical systems. Chapter 1 gives a general introduction into the sources of gene expression noise and non-genetic cell-to-cell variability. Throughout Chapters 2-5, we investigate cell-to-cell variability in a wide range of biological systems through the integration of experimental approaches and mathematical modelling.In Chapter 2, we derive a novel approximate method to obtain closed-form expressions for the means, variances, and power spectra of the molecular species within a chemical reaction network with intrinsic and slowly changing extrinsic noise. We do this by extending the conventional linear-noise approximation (LNA) to include systems where extrinsic noise manifests itself as fluctuations in the reaction rates. These extrinsic fluctuations are characterised by a longer timescale compared to the typically fast intrinsic reaction processes. We verify the accuracy of the theory by its application to different models of gene regulatory networks and comparing the analytical predictions to the more computationally costly results produced through stochastic simulations.In Chapter 3, we study the bacterial adaptive immune system CRISPR-Cas in a growing population of Escherichia coli. To survive infection, bacteria are forced to obtain an immunological memory of past infections through a process termed priming, to enable recognition of previously encountered threats. In addition, the cell contains an operon of CRISPR-associated (Cas) proteins which are responsible for finding and eliminating these invaders through a process called interference. We use time-lapse microscopy in combination with microfluidics to obtain single-cell lineage data throughout the entire duration of CRISPR defence. For the first time, we quantify the variation that exists between cells in how fast they can respond to foreign mobile genetic elements. Clearance of previously encountered invaders through CRISPR interference is fast with a narrow distribution. However, invaders can accumulate mutations in the PAM region which allows them to escape direct interference, resulting in large cell-to-cell variability of clearance times. Further analysis of the experimental data, together with a specially developed agent-based stochastic framework which simulates the behaviour of the bacterial population, allow us to identify the acquisition of a new immunological memory (adaptation) as the source of the increased variation in priming. Statistical analysis of cell lineage features reveals that faster growth and cell division, as well as higher levels of the CRISPR surveillance complex Cascade, increase the probability of plasmid clearance by interference. In contrast, slower growth is associated with a higher rate of adaptation. Through mathematical modelling we estimate the influence of target and Cascade copy numbers, as well as Cascade binding affinity of the rate of priming. Our results show that the ability to adapt to an invading threat by primed CRISPR adaptation is highly stochastic, implying that only subpopulations of bacteria can respond to foreign invaders in a timely manner.It has been shown that mutations in the protospacer adjacent motif (PAM), which flanks the DNA sequence targeted by the surveillance complex, affect target recognition by Cascade. In Chapter 3, we found that this reduced binding affinity of Cascade to targets with escaping PAM mutations can explain the observed wide distribution of plasmid loss times. However, the exact mechanism by which Cascade initiates primed adaptation, the acquisition of a new immunological memory, is largely unknown. In Chapter 4, we investigate the primed adaptation mechanism further by characterising the dynamics of CRISPR-mediated target clearance for three different PAM variants. We compare experimental single-cell lineages to simulated trajectories from two mechanisms of primed adaptation proposed in the literature by adapting the agent-based framework developed in the previous chapter. We show that features of the data are consistent with the interference-independent model for adaptation, in which a primed adaptation complex is responsible for the acquisition of new immunological memories. Our results indicate that the CRISPR-response of E. coli depends strongly on the PAM variant, which suggests these bacteria might employ a strategy to balance the relative rates of interference and the acquisition of new memories.While for bacteria the notion of cell-to-cell variability to enable bet-hedging strategies is sensible, for multicellular organisms reproducible and coordinated development might seem more important. Plants have evolved regulatory mechanisms to achieve specialised cell types and robust tissue growth. In Chapter 5 we quantify the noisiness of stochastic gene expression in Arabidopsis thaliana at the cellular level. To this end, we employ a combination of experimental and modelling approaches. First, we use the photoconvertible KikGR marker to show that the protein expressions of individual cells fluctuate over time. A dual reporter system is then used to study extrinsic and intrinsic noise. This reveals that extrinsic noise is the main source of protein variability in both young and old rosette leaves, and that extrinsic noise in stomata is clearly lower in comparison to several other cell types. Finally, through spatial analysis we show that cells are coupled with respect to stochastic protein expression in young leaves, hypocotyls and roots but not in mature leaves. Through theoretical analysis we find that the observed spatial correlation between cells can only partially be explained by the inheritance of mRNA and protein from a shared ancestor, which suggests other extrinsic noise sources affect the cellular dynamics.The results from Chapters 2-5 are discussed in a broader context in Chapter 6. In this general discussion, experimental and computational challenges relating to the study of gene expression noise are reviewed. Furthermore, I discuss possible implications of stochastic gene expression for phenotypic variability.

Advances in genome sequencing and high-throughput technologies have boosted the development of Synthetic biology and Systems biology. Synthetic biology aims to create and reprogram natural systems. Advances in Synthetic biology has facilitated the adoption of the design build test learn cycles into metabolic engineering. The DBTL cycles are a recursive loop that aims to optimize the development of microbial factories in a more systematic and efficient manner. Systems biology aims to study living organism at system level using holistic approaches. Among different modelling tools in Systems biology, genome-scale constraint-based metabolic model is the most successful approach to study the whole metabolic network. GEM is comprehensive knowledgebase that contain metabolic reactions that known to occur in a target organism. GEMs have been used in many applications to guide metabolic engineering and contextualizing ‘omics’ data. The objective of this thesis is to deploy GEMs for microbial cell factories and evaluate their main technical limitations.Chapter 1 describes two paradigms: reductionism and holism in life sciences, Systems biology, genome-scale constraint-based metabolic models, Synthetic biology and the design build test learn cycle. Chapter 1 provides the background for all other chapters.In Chapter 2 I constructed a GEM for Cutaneotrichosporon oleaginosus to model its lipid production. C. oleaginosus is a fast-growing oleaginous yeast that can grow in a wide range of low-cost carbon sources. I constructed a GEM to increase our understanding of this yeast and provide a knowledge base for further industrial use. A new modelling approach was introduced to account for changes in the biomass composition of this organism in conditions with high carbon to nitrogen (C/N) ratio in the media. This modelling approach is shown to better predict conditions with high lipid accumulation using glucose, fructose, sucrose, xylose, and glycerol as sole carbon source. The model suggests ATP-citrate lyase as a possible target to further improve lipid.Producing chemicals from living cells has been considering a sustainable approach to life. The biosynthesis of many natural compounds is still limited due to the lack of efficient synthesis routes. As a showcase of how GEMs can assist in designing pathways for chemical production in microbes, in Chapter 3 I employed GEMs to design and evaluate pathways for cis,cis-muconic acids, anisole, aniline, 3-methylmalate, and geranic acid production in Pseudomonas putida in the context of the Design-Build-Test-Learn cycles. I established a general system to rank these pathways based on thermodynamic feasibility, enzyme sequence availability, and maximum theoretical yield. Among our target compounds, cis,cis-muconic acid is a well-known chemical with many published biosynthesis pathways. For this well-studied target, we predicted 8 pathways, of which 2 pathways had not been reported earlier. Novel pathways were predicted to produce anisole, aniline, 3-methylmalate, and geranic acid.While constructing and using GEMs in Chapters 2 and 3, two main problems have recurred. The first problem is the use of inconsistent namespaces among GEMs. A critical step in constructing GEMs is to manually curate them by integrating information from independent (organism-specific) sources to provide a comprehensive representation of what is presently known about the metabolism of the modelled organism. Combining this precious information from individual GEMs to make a consensus model of the organism is essential. Using models from different species as a foundation to construct a new model can help to avoid repeating the same time-consuming manual curation step. In addition, GEMs need to be updated continuously since new knowledge is coming in short order. However, such simple tasks cannot be done easily due to a simple reason: inconsistent namespaces. GEMs constructed for different organisms by different researchers often use different naming conventions depending on which databases were selected for model construction. While mapping between namespaces seems like the only fair solution, it involves a high risk of mismatch and may invalidate the model. I evaluated the (in)consistency of names and non-systematic identifiers used in 11 biochemical databases of biochemical reactions and the problems that arise when mapping between different namespaces and databases in Chapter 4. I found that such inconsistencies can be as high as 83.1%, thus emphasizing the need for strategies to deal with these issues. Currently, manual verification of the mappings appears to be the only solution to remove inconsistencies when combining models.The second problem that has arisen relates to the efficiency of gap-filling tools. The lack of accurate functional annotations often renders GEMs incomplete, giving rise to missing reactions, the so-called ‘gaps’ in the network. Gap-filling becomes important during model construction not only to make a functional model but also to generate new knowledge on protein function. To assist gap-filling, many algorithms have been published. To be able to use GEMs effectively, these methods should allow the model to be as accurate as possible, preferably also in a user-friendly manner so that they become available to many researchers. However, gap-filling algorithms vastly differ in their objectives, implementation platforms, and input data requirements. These differences imply a variety in their usability and accuracy. In Chapter 5 I conducted an extensive evaluation of these algorithms from a user’s perspective. We found that most of the tools are not used due to the lack of a workable implementation. From those for which an implementation is readily available, we selected SMILEY, FASTGAPFILL and Meneco to further investigate their performances. SMILEY is the best among the three algorithms for small-scale degradation.Finally, in Chapter 6 I discussed the three significant themes stood out across Chapters 2,3,4,5 1) the lack of standards in namespaces, tool development, and guidelines for model evaluation, 2) the need to improve models and computational tools, for instance to account for uncertainty in the biomass synthesis reaction or to improve gap-filling algorithms, and 3) the potential contribution of GEMs to the DBTL cycle.In conclusion, the work presented in this thesis illustrates how the lack of standards in GEMs can hamper their potential. GEMs have great potential in the DBTL cycles. Standardization and improvement in GEM formulation are needed to maximize the use of these models.

Mycoplasmas are bacteria belonging to the Mollicutes class. They infect a wide range of organisms, among which humans, farm animals, herd animals and pets. Infections by mycoplasmas are currently treated with antibiotics, although many are ineffective due to the lack of cell wall of these bacteria. Moreover, this strategy leads to the development of antibiotics-resistant bacteria. The use of vaccines against mycoplasmas may be a solution for preventing wide spreading of the chronic infections they cause.The human pathogen Mycoplasma pneumoniae, causative agent of atypical pneumonia, can be used as a universal chassis to be deployed as a single- or multi-vaccine in a range of animal hosts. However, due to its reduced genome and dearth of many biosynthetic pathways, this bacterium depends on rich, undefined medium for growth, which makes the large-scale production of the vaccine challenging and expensive. For this reason, a genome-scale, constraint-based metabolic model was deployed to design a serum-free medium that supports growth of Mycoplasma pneumoniae with a rate comparable to the one obtained in rich medium. Model simulations highlighted, among the other components, the importance of two key lipids, part of Mycoplasma pneumoniae’s membrane. The serum-free medium MCMyco was then developed including these components and tested in vitro, showing robust growth of a range of mycoplasmas. Another strategy to increase Mycoplasma pneumoniae’s growth consisted in the design of a fatty acid-prototrophic strain of Mycoplasma. Indeed, despite the inability of synthetizing fatty acids, Mycoplasma pneumoniae needs them to construct its membrane lipids. The study is performed through a combination of DNA engineering, genome-scale modelling and gap-filling algorithm approaches, revealing a bottleneck related to NADPH, one the cofactor mostly used in the fatty acid biosynthesis pathway. As a result, the computationally-designed strain implements a pathway synthetizing fatty acids, which are then integrated as acyl chains into the membrane lipids, and genetic modifications leading to a more efficient NADPH regeneration.The deployment of Mycoplasma pneumoniae as a universal chassis for vaccination might lead to issues related to the causative link of the bacterium to post-infectious Guillain-Barré-Stohl syndrome, a neurodegenerative disorder. Galactocerebroside, a glycolipid synthetized by Mycoplasma, has been identified as the compound that triggers the autoimmune reactions causing the syndrome. The biosynthesis pathway of this lipid was therefore investigated, with the aim of finding those enzymes of Mycoplasma pneumoniae to target to prevent the synthesis. The computational analysis, performed through genomic comparison, functional annotation and Hidden Markov Models, was extended to all mycoplasmas, revealing not all these bacteria have the potential of synthetizing galactocerebroside. In consequence, such group of mycoplasmas could be considered safe for application. A protein motif that, when found in the enzyme synthetizing the glycolipid, could be an indication of synthesis capability in mycoplasmas was built. This motif can be used it as a signature to investigate on the potentiality of galactocerebroside synthesis in microorganisms.A universal Mycoplasma pneumoniae-based vaccine is a genetically-modified organism produced at large-scale and deployed in open environment. Therefore, biocontainment strategies must be applied to prevent its proliferation outside the desired operating conditions. Several Ordinary Differential Equations models were developed to sustain the experimental design of biosafety circuits to insert into the Mycoplasma pneumoniae’s genome. The model yielded robust results with a killing switch inserted in the bacterial genome in double copy, which showed to be resistant to inactivating mutations over several strain passages.The dissertation ends with a discussion on the computational approaches used, a recapitulation of the status of the vaccine and its future perspectives. Moreover, the importance of serum-free media in growth cultures is highlighted, with a special focus on their model-driven design. A reflection on the safety of live attenuated vaccine is proposed, and, finally, safety-by-design is suggested as a strategy to be implemented in the developmental phase of therapeutic and pharmaceutical products.

Using Pseudomonas putida KT2440 as a proof-of-concept organism, this thesis was aimed at microbial lifestyle engineering for industrial applications. In this thesis, a structured approach was applied by first determining what microbial improvements industry is looking for by conducting a series of interviews with both industry and academia. Besides pinpointing the fields of interest from an industrial perspective, the interviews also clarified the limitations of the actual implementation of novel or (synthetically) adapted strains developed. Strain safety being at the top of their list, we first checked the claimed GRAS safety level of P. putida KT2440.A major obstacle for the breakthrough of P. putida KT2440 to be widely used as a biotechnological host is its obligate aerobic metabolism. In silico-directed strain improvement were initiated by the adaptation of strict aerobic P. putida KT2440 to micro-oxic and anoxic conditions. Adaptation to micro-oxic levels was done by first creating a design for a recombinant strain capable of anaerobic fermentation. The bottlenecks uncovered were resolved by insertion of three genes, and the recombinant strains were monitored through an adaptive laboratory evolution method with oxygen gradients set up specifically for this purpose. Recombinant strains were able to grow under micro-oxic conditions. Strain performance did not improve compared to the negative control under anoxic conditions. A more elaborate in-silico analysis was performed, combining protein domain analysis, transcriptomic analysis and genome-scale metabolic models to design a recombinant P. putida KT2440 strain capable of anaerobic respiration.Another general limitation in strains is their limited thermo-tolerance. We discovered a strong universal connection between NAD+ availability and thermo-tolerance. By replacing one single gene for a thermophilic heterolog in mesophilic prokaryotes, both P. putida and E. coli showed instant improved thermo-tolerance. Insertion of the aspartate NAD+ biogeneration pathway in eukaryotic yeast S. cerevisiae resulted in a similar effect. To determine the value of this thermo-tolerance in industry, a down-scaled microfluidics system was developed to mimic temperature fluctuations occurring in large scale bioreactors. The novel discovery between thermo-tolerance and NAD+ availabilty was patented.

Microalgae are unicellular microscopic and photosynthetic organisms. They are found all over the planet and in all sorts of environments. Their role has been and is still very important for the planet, most notably that they are currently accounted for producing half of the atmospheric oxygen. While they used to be studied because of their capabilities to depollute water, the interests have shifted towards oleaginous microalgae and their high level of fatty-acids accumulation. Fatty-acids such as Triacylglycerides (TAG) are of particular interest for their easy chemical treatment to produce clean biodiesel. Even if microalgae have higher energy conversion efficiency than plants, do not need arable lands to grow on, and do not compete with feed production, the optimal conditions of production are still too costly to compete with fossil fuel pricing. To decrease production costs, growth conditions and the physiological efficiency of the microalgae needs to be optimized. This requires a deep understanding of the microalgae phenotype in the relevant growth conditions. A phenotype represents a set of behavioral traits of an organism In this thesis, the internal phenotype, the transcriptional landscape of two oleaginous microalgae species was studied using different growth conditions. The RNA content was chosen because it provides a dynamic system-wide view, can be done in high-throughput and, as proxy, can inform about the cellular and metabolic activities in response to a changing environment.To analyze the transcriptome, it is necessary to know the functions of the transcripts. In contrast to model organisms like human or Arabidopsis thaliana whose genomes have been deeply annotated and studied, microalgae are far from that state. For both organisms studied in this thesis, Neochloris oleoabundans and Tetradesmus obliquus, it was necessary to annotate the genes and transcripts since it was never done before. To functionally annotate a gene, most methods rely on sequence similarities to identify the closest gene in known organisms. Green algae are a difficult case due to the large genetic distance between them and the greater distance to reference organisms from land plants. Chapter 2 treats about the particular problem of protein annotation in microalgae. I analyzed the general state of data availability in microalgae, and we discussed about several annotation methods that are better suited than sequence similarity and discussed the limitations of using domain-based recognition methods. Besides, we also discussed the identification of the protein localization within the different subcellular compartment in microalgae. Finally, we stressed the importance of a large-scale wet-lab efforts for a few selected microalgae in order to provide a solid foundation for computation based methods.To obtain more insight into the metabolism of Ettlia oleoabundans in Chapter 3 , a constraint-based metabolic model of Ettlia oleoabundans was built around the central carbon metabolism. This model was built based on the knowledge of central metabolism of algae at that moment and was cross checked with the de novo assembled annotated transcriptome. Experiments in controlled turbidostat were conducted in different combinations of light intensity and nitrogen supply. The measurements from the experimental conditions were used as constraints on the inputs and outputs of the model, effectively allowing us to estimate the metabolic flux distributions. In addition, RNA samples from the different experimental conditions were sequenced and analyzed. These data were used to validate the model structure as stated before, correlate expression levels with flux distributions and get a better understanding of the effect of light and nutrient conditions on algal physiology. The metabolic model calculates a maximum TAG yield of N. oleoabundans on light of 1.06 g (mol photons)-1, more than 3 times the current experimental yield under optimal conditions. The model also shows that TAG yield on light can be more efficiently improved by optimizing photosynthetic conversion than by blocking competing pathways. Geranylgeranyl diphosphate reductase was identified as a potential regulator for photosynthetic capability that complements the fine-tuning of chlorophyll levels from synthesis and degradation. Finally, we identified some key reactions that could be targeted to improve TAG yield, by not only specifically increasing the flux within the lipids and TAG pathways, but also potentially redirect carbons from other pathways.Water is a precious resource, and using fresh drinkable water to grow plants or microalgae could be considered non sustainable. However, salt-water is abundantly available and would be cheaper to supply. Therefore it is important to understand how algae deal with a salt water environment under growth and production conditions. This allowed us in Chapter 4 to study how algae deal with high salinity conditions under nitrogen replete (growth) and nitrogen deplete (TAG accumulation) conditions using a transcriptomics approach. The oleaginous microalgae, Ettlia oleoabundans (formally known as Neochloris oleoabundans) was chosen as a model algae, since it can accumulate large amounts of TAG and can grow in both fresh and salt-water. For this algae experiments were done in fresh water and salt water in combination with nitrogen replete and nitrogen deplete conditions. In addition to the transcriptome, we analyzed the biomass composition including TAG and starch accumulation and used the data to look into different salt resistance mechanisms. We found that Proline and the ascorbate-glutathione cycle seem to be of importance for successful osmoregulation in N. oleoabundans. Genes involved in Proline biosynthesis were found to be upregulated in salt water, which is supported by Nuclear magnetic resonance (NMR) spectroscopy. Oil accumulation is increased under nitrogen-deplete conditions in a comparable way in both fresh and salt water. The mechanism behind the biosynthesis of compatible osmolytes can be used to improve N. oleoabundans and other industrially relevant microalgal strains to create a more robust and sustainable production platform for microalgae derived products in the future.Although the TAG content that can be reached in Ettlia oleoabundans is high, the volumetric TAG productivity in Tetradesmus obliquus was evaluated to be clearly higher, while reaching the same TAG content. This was mainly due to the ability of T. obliquus to maintain photosynthetic efficiency for a longer time longer during the nitrogen depletion phase. Therefore, it was decided to switch to T. obliquus as a model organism. To obtain an idea of the capabilities of T. obliquus and to make transcriptome experiments easier to analyze, the genome of T. obliquus was sequenced. In Chapter 5 , the sequencing of the genome of T. obliquus is presented. The assembly approach was unconventional by combining two different methods and was able to combine the higher coverage from one method with the precision of the other method. In this way, the coverage and the accuracy of the assembly was maximized.Production using microalgae will in many cases occur outdoors using sunlight. Consequently the algae will be exposed to the naturally occurring day night cycle. To better understand the effect of these day night cycles, in Chapters 6 and 7, the transcriptional response of algae to diurnal cycles was studied under nitrogen replete conditions and nitrogen limiting conditions for the wild type and a mutant that can not synthesize starch. In Chapter 6 , hourly samples of RNA of Tetradesmus obliquus UTEX 393 were taken from a turbidostat culture operated under nitrogen replete conditions over a diurnal cycle of 16 hour light and 8 hours dark, to obtain more insight in in the transcriptional response towards diurnal cycles. In addition, to understand the effects of a lack of starch, the major transient energy storage, we sequenced samples of the starchless mutant slm1 that were collected every three hours under the same conditions. At the same time, samples were collected and measurements of the biochemical composition of biomass and the specific light absorption rate were performed. These data are presented in a previously published article [24]⁠. The genome features were annotated using more than 38 RNA-seq samples from this study, using a specially developed extension of the FAIR principle based framework called SAPP. The work done to extend this framework for transcriptome analysis is described in the discussion Chapter 8 . We described the succession of metabolic events that occurred during the diurnal cycle, which are in agreement with the biochemical measurements. Comparing the wild-type with the starchless mutant slm1, we found a few temporal shifts in expression that reflect transcriptional adaptation to the lack of a transient energy storage compound during the dark period. Our study provides new perspectives on the role of starch and the adaptation to LD cycles of oleaginous microalgae.In Chapter 7 a similar experimental approach was taken, where samples were taken for biochemical and transcriptome analysis every 3 hours from a turbidostat culture operated at the same diurnal cycle of 16 hours light and 8 hours dark, but this time in nitrogen limiting conditions, resulting in TAG accumulation. Again this was done for the wild type and the starchless mutant. The transcriptional landscape and biochemical data are compared to the nitrogen replete condition in Chapter 6 , to evaluate the effect of nitrogen limitation in general and study how the lack of starch is affecting TAG accumulation under nitrogen limiting conditions. We observed that the cycling diurnal effect is greatly reduced in comparison to nitrogen replete condition. The wild-type accumulated more starch than in nitrogen replete condition (Chapter 6 ), and small oscillation was observed, indicating that it is being used as transient energy storage during the dark period. While the biochemical analysis did not reveal any oscillation in total lipids content in either strain, slm1 over-expresses transcripts associated to TAG and lipid degradation during the dark period. Besides, while slm1 accumulated more TAG than the wild-type, its conversion efficiency was only half of the wild-type. It appears also that the organism recycles more proteins during the dark period to supply nitrogen for the strong increase in amino acid synthesis right after light is turned on.In Chapter 8 , the results of this thesis are discussed. The interest in analyzing the transcriptome of microalgae is explained. Then, the annotation methods are explained to show the large improvements between them, and could it still be improved. Basic numbers of the later annotation results are compared to the current status in UniProtKB. Particular points from the transcriptomics data are discussed, notably the expression of the mutant gene responsible for slm1 phenotype, and the interests from using single-cell technologies. Suggestions are then made to improve the experimental conditions and the photobioreactors setups. The efforts made in the thesis to generate and store the data according to the FAIR principles are explained. Finally, using the knowledge acquired during this thesis, suggestions are made to improve the growth conditions and to improve TAG production with divers metabolic engineering strategies. The work of this thesis contributes for the future of sustainable production of biofuels, which ultimately will help alleviating human society’s dependence on fossil fuels.

The goal of this thesis is to improve the prediction of genotype to phenotypeassociations with a focus on metabolic phenotypes of prokaryotes. This goal isachieved through data integration, which in turn required the development ofsupporting solutions based on semantic web technologies. Chapter 1 providesan introduction to the challenges associated to data integration. Semantic webtechnologies provide solutions to some of these challenges and the basics ofthese technologies are explained in the Introduction. Furthermore, the ba-sics of constraint based metabolic modeling and construction of genome scalemodels (GEM) are also provided. The chapters in the thesis are separated inthree related topics: chapters 2, 3 and 4 focus on data integration based onheterogeneous networks and their application to the human pathogen M. tu-berculosis; chapters 5, 6, 7, 8 and 9 focus on the semantic web based solutionsto genome annotation and applications thereof; and chapter 10 focus on thefinal goal to associate genotypes to phenotypes using GEMs. Chapter 2 provides the prototype of a workflow to efficiently analyze in-formation generated by different inference and prediction methods. This me-thod relies on providing the user the means to simultaneously visualize andanalyze the coexisting networks generated by different algorithms, heteroge-neous data sets, and a suite of analysis tools. As a show case, we have ana-lyzed the gene co-expression networks of M. tuberculosis generated using over600 expression experiments. Hereby we gained new knowledge about theregulation of the DNA repair, dormancy, iron uptake and zinc uptake sys-tems. Furthermore, it enabled us to develop a pipeline to integrate ChIP-seqdat and a tool to uncover multiple regulatory layers. In chapter 3 the prototype presented in chapter 2 is further developedinto the Synchronous Network Data Integration (SyNDI) framework, whichis based on Cytoscape and Galaxy. The functionality and usability of theframework is highlighted with three biological examples. We analyzed thedistinct connectivity of plasma metabolites in networks associated with highor low latent cardiovascular disease risk. We obtained deeper insights froma few similar inflammatory response pathways in Staphylococcus aureus infec-tion common to human and mouse. We identified not yet reported regulatorymotifs associated with transcriptional adaptations of M. tuberculosis.In chapter 4 we present a review providing a systems level overview ofthe molecular and cellular components involved in divalent metal homeosta-sis and their role in regulating the three main virulence strategies of M. tu-berculosis: immune modulation, dormancy and phagosome escape. With theuse of the tools presented in chapter 2 and 3 we identified a single regulatorycascade for these three virulence strategies that respond to limited availabilityof divalent metals in the phagosome. The tools presented in chapter 2 and 3 achieve data integration throughthe use of multiple similarity, coexistence, coexpression and interaction geneand protein networks. However, the presented tools cannot store additional(genome) annotations. Therefore, we applied semantic web technologies tostore and integrate heterogeneous annotation data sets. An increasing num-ber of widely used biological resources are already available in the RDF datamodel. There are however, no tools available that provide structural overviewsof these resources. Such structural overviews are essential to efficiently querythese resources and to assess their structural integrity and design. There-fore, in chapter 5, I present RDF2Graph, a tool that automatically recoversthe structure of an RDF resource. The generated overview enables users tocreate complex queries on these resources and to structurally validate newlycreated resources. Direct functional comparison support genotype to phenotype predictions.A prerequisite for a direct functional comparison is consistent annotation ofthe genetic elements with evidence statements. However, the standard struc-tured formats used by the public sequence databases to present genome an-notations provide limited support for data mining, hampering comparativeanalyses at large scale. To enable interoperability of genome annotations fordata mining application, we have developed the Genome Biology OntologyLanguage (GBOL) and associated infrastructure (GBOL stack), which is pre-sented in chapter 6. GBOL is provenance aware and thus provides a consistentrepresentation of functional genome annotations linked to the provenance.The provenance of a genome annotation describes the contextual details andderivation history of the process that resulted in the annotation. GBOL is mod-ular in design, extensible and linked to existing ontologies. The GBOL stackof supporting tools enforces consistency within and between the GBOL defi-nitions in the ontology. Based on GBOL, we developed the genome annotation pipeline SAPP (Se-mantic Annotation Platform with Provenance) presented in chapter 7. SAPPautomatically predicts, tracks and stores structural and functional annotationsand associated dataset- and element-wise provenance in a Linked Data for-mat, thereby enabling information mining and retrieval with Semantic Webtechnologies. This greatly reduces the administrative burden of handling mul-tiple analysis tools and versions thereof and facilitates multi-level large scalecomparative analysis. In turn this can be used to make genotype to phenotypepredictions. The development of GBOL and SAPP was done simultaneously. Duringthe development we realized that we had to constantly validated the data ex-ported to RDF to ensure coherence with the ontology. This was an extremelytime consuming process and prone to error, therefore we developed the Em-pusa code generator. Empusa is presented in chapter 8. SAPP has been successfully used to annotate 432 sequenced Pseudomonas strains and integrate the resulting annotation in a large scale functional com-parison using protein domains. This comparison is presented in chapter 9.Additionally, data from six metabolic models, nearly a thousand transcrip-tome measurements and four large scale transposon mutagenesis experimentswere integrated with the genome annotations. In this way, we linked gene es-sentiality, persistence and expression variability. This gave us insight into thediversity, versatility and evolutionary history of the Pseudomonas genus, whichcontains some important pathogens as well some useful species for bioengi-neering and bioremediation purposes. Genome annotation can be used to create GEM, which can be used to betterlink genotypes to phenotypes. Bio-Growmatch, presented in chapter 10, istool that can automatically suggest modification to improve a GEM based onphenotype data. Thereby integrating growth data into the complete processof modelling the metabolism of an organism. Chapter 11 presents a general discussion on how the chapters contributedthe central goal. After which I discuss provenance requirements for data reuseand integration. I further discuss how this can be used to further improveknowledge generation. The acquired knowledge could, in turn, be used to de-sign new experiments. The principles of the dry-lab cycle and how semantictechnologies can contribute to establish these cycles are discussed in chapter11. Finally a discussion is presented on how to apply these principles to im-prove the creation and usability of GEM’s.

The aim of this thesis was to increase our understanding on how genome information leads to function and phenotype. To address these questions, I developed a semantic systems biology framework capable of extracting knowledge, biological concepts and emergent system properties, from a vast array of publicly available genome information. In chapter 2, Empusa is described as an infrastructure that bridges the gap between the intended and actual content of a database. This infrastructure was used in chapters 3 and 4 to develop the framework. Chapter 3 describes the development of the Genome Biology Ontology Language and the GBOL stack of supporting tools enforcing consistency within and between the GBOL definitions in the ontology (OWL) and the Shape Expressions (ShEx) language describing the graph structure. A practical implementation of a semantic systems biology framework for FAIR (de novo) genome annotation is provided in chapter 4. The semantic framework and genome annotation tool described in this chapter has been used throughout this thesis to consistently, structurally and functionally annotate and mine microbial genomes used in chapter 5-10. In chapter 5, we introduced how the concept of protein domains and corresponding architectures can be used in comparative functional genomics to provide for a fast, efficient and scalable alternative to sequence-based methods. This allowed us to effectively compare and identify functional variations between hundreds to thousands of genomes. In chapter 6, we used 432 available complete Pseudomonas genomes to study the relationship between domain essentiality and persistence. In this chapter the focus was mainly on domains involved in metabolic functions. The metabolic domain space was explored for domain essentiality and persistence through the integration of heterogeneous data sources including six published metabolic models, a vast gene expression repository and transposon data. In chapter 7, the correlation between the expected and observed genotypes was explored using 16S-rRNA phylogeny and protein domain class content as input. In this chapter it was shown that domain class content yields a higher resolution in comparison to 16S-rRNA when analysing evolutionary distances. Using protein domain classes, we also were able to identify signifying domains, which may have important roles in shaping a species. To demonstrate the use of semantic systems biology workflows in a biotechnological setting we expanded the resource with more than 80.000 bacterial genomes. The genomic information of this resource was mined using a top down approach to identify strains having the trait for 1,3-propanediol production. This resulted in the molecular identification of 49 new species. In addition, we also experimentally verified that 4 species were capable of producing 1,3-propanediol. As discussed in chapter 10, the here developed semantic systems biology workflows were successfully applied in the discovery of key elements in symbiotic relationships, to improve functional genome annotation and in comparative genomics studies. Wet/dry-lab collaboration was often at the basis of the obtained results. The success of the collaboration between the wet and dry field, prompted me to develop an undergraduate course in which the concept of the “Moist” workflow was introduced (Chapter 9).

Approximately one-fourth of the Mycobacterium tuberculosis (Mtb) genome contains genes that encode enzymes directly involved in its metabolism. These enzymes represent potential drug targets that can be systematically probed with constraint based (CB) models through the prediction of genes essential (or the combination thereof) for the pathogen to grow. However, gene essentiality depends on the growth conditions and, so far, no in vitro model precisely mimics the host at the different stages of mycobacterial infection, limiting model predictions. A first step in creating such a model is a thoroughly curated and extended genome-scale CB metabolic model of Mtb metabolism. The history of genome-scale CB models of Mtb metabolism up to model sMtb are discussed and sMtb is quantitatively validated using 13C measurements. The human pathogen Mtb has the capacity to escape eradication by professional phagocytes. During infection, Mtb resists the harsh environment of phagosomes and actively manipulates macrophages and dendritic cells to ensure prolonged intracellular survival. In contrast to many other intracellular pathogens, it has remained difficult to capture the transcriptome of mycobacteria during infection due to an unfavorable host-to-pathogen ratio. The human macrophage-like cell line THP-1 was infected with the attenuated Mtb surrogate Mycobacterium bovis Bacillus Calmette–Guerin (M. bovis BCG). Mycobacterial RNA was up to 1000-fold underrepresented in total RNA preparations of infected host cells. By combining microbial enrichment with specific ribosomal RNA depletion the transcriptional responses of host and pathogen during infection were simultaneously analyzed using dual RNA sequencing. Mycobacterial pathways for cholesterol degradation and iron acquisition are upregulated during infection. In addition, genes involved in the methylcitrate cycle, aspartate metabolism and recycling of mycolic acids are induced. In response to M. bovis BCG infection, host cells upregulate de novo cholesterol biosynthesis presumably to compensate for the loss of this metabolite by bacterial catabolism. By systematically probing the metabolic network underpinning sMtb, the reactions that are essential for Mtb are identified. A majority of these reactions are catalyzed by enzymes and thus represent candidate drug targets to fight an Mtb infection. Modeling the behavior of the bacteria during infection requires knowledge of the so-called biomass reaction that represents bacterial biomass composition. This composition varies in different environments or bacterial growth phases. Accurate modeling of all fluxes through metabolism under a given condition at a moment in time, the so called metabolic state, requires a precise description of the biomass reaction for the described condition. The transcript abundance data obtained by dual RNA sequencing was used to develop a straightforward and systematic method to obtain a condition-specific biomass reaction for Mtb during in vitro growth and during infection of its host. The method described herein is virtually free of any pre-set assumptions on uptake rates of nutrients, making it suitable for exploring environments with limited accessibility. The condition-specific biomass reaction represents the 'metabolic objective' of Mtb in a given environment (in-host growth and growth on defined medium) at a specific time point, and as such allows modeling the bacterial metabolic state in these environments. Five different biomass reactions were used predict nutrient uptake rates and gene essentiality. Predictions were subsequently compared to available experimental data. Nutrient uptake can accurately be predicted, but accurate gene essentiality predictions remain difficult to obtain. By combining sMtb and a model of human metabolism, model sMtb-RECON was developed and used to predict the metabolic state of Mtb during infection of the host. Amino acids are predicted to be used for energy production as well as biomass formation. Subsequently the effect of increasing dosages of drugs, targeting metabolism, on the metabolic state of the pathogen was assessed and resulting metabolic adaptations and flux rerouting through various pathways is predicted. In particular, the TCA cycle becomes more important upon drug application, as well as alanine, aspartate, glutamate, proline, arginine and porphyrin metabolism, while glycine, serine and threonine metabolism become less important for survival. Notably, an effect of eight out of eleven metabolically active drugs could be recreated and two major profiles of the metabolic state were predicted. The profiles of the metabolic states of Mtb affected by the drugs BTZ043, cycloserine and its derivative terizidone, ethambutol, ethionamide, propionamide, and isoniazid were very similar, while TMC207 is predicted to have quite a different effect on metabolism as it inhibits ATP synthase and therefore indirectly interferes with a multitude of metabolic pathways.

The filamentous fungus Aspergillus niger is well-known as an industrial producer of a number of organic acids. However, despite decades of research, there are still bottlenecks and, with respect to the molecular mechanism underlying A. niger organic acids production there are still a number of unclarities. This thesis aims to gain a better understanding of the molecular mechanisms underlying organic acid production in A. niger and is mostly focused on active transport of carbon sources, organic acids and reactive oxygen species via transmembrane transport proteins. Additionally, the regulation of glucose oxidase expression system in gluconate pathways as well as response of A. niger to environmental factors during organic acid fermentation were also studied.

Itaconic acid (2-methylidenebutanedioic acid) is an important building block chemical that can be produced from sugars via chemical or biological conversions. Itaconic acid can be processed into a polymer, which can subsequently be used to replace the petroleum-based polyacrylic acids. Itaconic acid is naturally produced by Aspergillus terreus, certain Ustilago and Candida species and Pseudozyma antarctica. Also in mammalian cells itaconic acid is found during macrophage activation. Aspergillus niger is a filamentous fungus that is the current host of choice for the production of citric acid on an industrial scale, whereas A. terreus is the main natural producer of itaconic acid. The biosynthesis of itaconic acid is very similar to the biosynthesis of citric acid in A. niger. Citric acid concentrations of 200 g/L are nowadays obtainable with A. niger. Theoretically, this allows itaconic acid concentrations of over 135 g/L to be obtained with A. niger, which is higher that the concentrations currently obtained with A. terreus (80 g/L). However, in A. niger, cis-aconitate decarboxylase (CadA), the key enzyme for itaconic acid biosynthesis, is missing. Expression of A. terreus CadA in A. niger results in the production of a low itaconic acid concentration (0.05 g/L). The itaconic acid concentration was increased by the expression of a putative mitochondrial transporter (MttA) and a putative plasmamembrane transporter (MfsA). Expression of the MttA transporter in an itaconic acid producing A. niger strain resulted in a twenty-fold increase in itaconic acid secretion. Expression of the A. terreus itaconic acid cluster consisting of the cadA gene, the mttA gene and the mfsA gene resulted in A. niger strains that produce over twenty five-fold higher levels of itaconic acid and show a twenty-fold increase in yield compared to a strain expressing only CadA. To further increase itaconic acid production a modified 6-phosphofructo-1-kinase, pfkA, was expressed in a citrate producing A. niger strain in combination with cis-aconitate decarboxylase, cadA, from A. terreus. The combined expression of pfkA and cadA resulted in increased citrate levels, but did not show increased itaconic acid levels. The combined expression of pfkA with the itaconic acid biosynthetic cluster resulted in significantly increased itaconic acid production at earlier time points. Also the itaconic acid productivity increased significantly. The maximum itaconic acid productivity that was reached under these conditions equaled 0.15 g/L/h, which is only a factor 17 lower than the 2.5 g/L/h that, according to the US Department of Energy, should be achieved to have an economically feasible production process. With the requirement of an increasing number of different genes that needed to be integrated in the genome of A. niger, there appeared a need for an efficient flexible pathway transfer system. The developed Funbrick system enables easy and quick construction of expression vectors for integration of pathways and the construction of localization vectors. A major advantage of this system is the ability to freely alter the so- called Funbrick after its construction. The Funbrick system was successfully applied by integration of the itaconic acid biosynthesis gene cluster of A. terreus in the genome of A. niger. The highest itaconic acid concentration obtained equaled 169 mg/L. In addition, the Funbrick system was used to localize the itaconic acid biosynthesis proteins. In order to further increase the efficiency of pathway transfer in A. niger and other filamentous fungi, we developed a non-laborious way for the rapid selection of transformants resulting from integration at a defined locus in the genome. In A. niger, homologous or targeted integration is relatively rare. The A. niger genome contains a spore color formation gene fwnA, which was investigated as a selective marker for homologous integration. Transformants resulting from homologous recombination at the fwnA locus can be rapidly distinguished on the basis of their spore color, which is fawn. Overall, itaconic acid production has been successfully established in A. niger, albeit in low concentrations, by expressing the gene encoding the key enzyme CadA. Expression of the transporters MttA and MfsA from A. terreus in A. niger showed that especially the mitochondrial MttA transporter has a major effect on the production of itaconic acid in the host strain A. niger NW186. Overall, transport processes play a major role in itaconic acid production and the focus thereon has led to large improvements in itaconic acid production by A. niger.

Chapter 1 provides the background for the thesis. In it, I provide a short overview of what biotechnology is and how it has been utilized it for thousands of years. I then address how modern biotechnology has evolved over the past few decades. Its progress has been triggered by the discovery of nucleic acids and marked by a focus on genetic understanding of cell and organism function and on the subsequent manipulation to ultimately benefit society in one way or another. Furthermore, computational biology has been increasingly important to determine the success of biotechnological research, in for example an anti-malarial drug- producing yeast. However, for microalgae, which are very promising organisms for biotechnological applications, there are essentially no successful commercialized examples of modern biotechnology. The chapter further discusses the importance of computationally predicting protein functions and its role in bioinformatics and systems biology research, concluding that this is one of the challenges for microalgal biotechnology. This topic is discussed in all chapters of this thesis, as its overarching goal is to develop and deploy tools and methodologies that lead to increase our understanding of microalgae as cell factories.Chapter 2 is an review on the state of microalgal biotechnology in 2014, of which the major discussion points are still valid, and how bioinformatics and systems biology should be used to further microalgal research. It describes the challenges of microalgal genomics, bioinformatics, and systems biology research. The chapter addresses a few challenges for microalgae in particular: a lack of genomic data, a low amount of validated protein functions, and genome-scale metabolic models largely based off of Arabidopsis thaliana. Suggestions are made on how to overcome these challenges, by for example better utilizing bioinformatics methods and databases. Chapter 3 addresses a specific challenge: the need for accurate annotation of the functions of microalgal proteins. It exposes the lack of understanding we have of their protein functions, with a staggering 90% of their annotations also present in the distantly related plant Arabidopsis thaliana. Finally, this chapter outlines areas in which microalgal protein function prediction can be improved. In Chapter 4, I present CrowdGO, a prediction tool based on the “wisdom of the crowd” principle for protein function prediction that aims to overcome the major problem highlighted in Chapter 3. It operates by taking and merging the existing predictions made by other methods. These merged predictions are then put through a machine learning algorithm which is trained to recognize patterns in these predictions and correlate them to true or false positives. CrowdGO shows significantly higher accuracy, with a p-value < 2.22e-16, over existing prediction methods, as well as an improved precision and recall optimum.In Chapter 5 deploy CrowdGO to the genomics of the oleaginous yeast Cutaneotrichosporon curvatus, which thus serves as a real biological test case for the method. Comparisons between the CrowdGO annotated C. curvatus proteins to the existing ones of a related yeast showcases the potential of CrowdGO. GO enrichment analysis of C. curvatus between transcriptomes of normal growth conditions and nitrogen starved conditions shows cell maintenance functions enriched during the first, and stress functions enriched during the latter. This is in line with what one would expect for an oleaginous eukaryote and reassures us that the CrowdGO annotations are reliable. The CrowdGO annotations are further used in a manual annotation pipeline, which we used to manually curate over 700 metabolic C. curvatus proteins. These are used together with differential expression analysis to characterize triacylglycerol synthesis during nitrogen starvation conditions. Only one enzyme was missing after the first round of annotations, displaying a high recall for enzymes when using the manual annotation pipeline.In chapter 6 we study the comparative genomics between different Botryococcus braunii strains, an oleaginous eukaryote that either makes large amount of polysaccharides or hydrocarbons based on the strain. In this chapter, all methodologies discussed or developed in the previous chapters are used to try and identify the key genetic differences between the two strains that lead to polysaccharide or hydrocarbon synthesis. We use CrowdGO to annotate all the proteins, and perform manual annotation on a thousand metabolic proteins. These are used in conjunction with quantitative proteomics analysis of several conditions including light and dark, different nitrogen levels, and different cell phases. By combining the manual annotations and the proteomics analysis, we were able to characterize several key pathways including the non-mevalonate pathway, fucose synthesis pathway, and the TCA cycle. Analysis of these pathways reveals key differences in the expression of enzymes that are likely to correspond to polysaccharide or hydrocarbon synthesis. Apart from revealing some key features about Botryococcus braunii, this chapter serves as a template for future large-scale microalgal research.Chapter 7 is a general discussion on the thesis. In it, I discuss how the work in this thesis relates to the SPLASH project for microalgae. Furthermore, I discuss how microalgal annotations can still be improved through the use of various stages of bioinformatics, systems biology, and synthetic / metabolic engineering research. Finally, I discuss how microalgae have potential as protein farms, and how it might be possible to unlock this potential.

The gastrointestinal tract is subject of much research for its role in an organism’s health owing to its role as gatekeeper. The tissue acts as a barrier to keep out harmful substances like pathogens and toxins while absorbing nutrients that arise from the digestion of dietary components in in the lumen. There is a large population of microbiota that plays an important role in the functioning of the gut. All these sub-systems of the gastrointestinal tract contribute to the normal functioning of the gut. Due to its various functionalities, the gut is able to respond to different types of stimuli and bring the system back to homeostasis after perturbations. The work done in this thesis uses several bioinformatic tools to improve our understanding of the functioning of the gut. This was achieved with data from model animals, mice and pigs which were subjected to changing environments before their gastrointestinal response was measured. Different types of stimuli were studied (eg, antibiotic exposure, changing diets and infection with pathogens) in order to understand the response of the gut to varying environments. This data was analysed using different data integration techniques that provide a holistic view of the gut response. Vertical data integration techniques look for associations between different types of ~omics data to highlight possible interactions between the measured variables. Lateral integration techniques allow the study of one type of ~omics data over several time points or several experimental conditions. Using these techniques, we show proof of interactions between different sub-systems of the gut and the functional plasticity of the gut. Of the several hypotheses generated in this thesis we have validated several using existing literature and one using an in-vitro system. Further validation of these hypotheses will increase understanding of the responses of the gut and the interactions involved.

The goals of this thesis are to increase the understanding of microbial metabolism and to functionally (re-)design microbial systems using Genome- Scale Metabolic models (GSMs). GSMs are species-specific knowledge repositories that can be used to predict metabolic activities for wildtype and genetically modified organisms. Chapter 1 describes the assumptions associated with GSMs, the GSM generation process, common GSM analysis methods, and GSM-driven strain design methods. Thereby, chapter 1 provides a background for all other chapters. In this work, there is a focus on the metabolically versatile bacterium Pseudomonas putida (chapters 2,3,4,5,6), but also other model microbes and biotechnologically or societally relevant microbes are considered (chapters 3,4,6,7,8).GSMs are reflections of the genome annotation of the corresponding organism. For P. putida, the genome annotation that GSMs have been built on is more than ten years old. In chapter 2, this genome annotation was updated both on a structural and functional level using state-of-the-art annotation tools. A crucial part of the functional annotation relied on the most comprehensive P. putida GSM to date. This GSM was used to identify knowledge gaps in P. putida metabolism by determining the inconsistencies between its growth predictions and experimental measurements. Inconsistencies were found for 120 compounds that could be degraded by P. putida in vitro but not in silico. These compounds formed the basis for a targeted manual annotation process. Ultimately, suitable degradation pathways were identified for 86/120 as part of the functional reannotation of the P. putida genome.For P. putida there are 3 independently generated GSMs, which is not uncommon for model organisms. These GSMs differ in generation procedure and represent different and complementary subsets of the knowledge on the metabolism of the organism. However, the differing generation procedures also makes it extremely cumbersome to compare their contents, let alone to combine them into a single consensus GSM. Chapter 3 addresses this issue through the introduction of a computational tool for COnsensus Metabolic Model GENeration (COMMGEN). COMMGEN automatically identifies inconsistencies between independently generated GSMs and semi-automatically resolves them. Thereby, it greatly facilitates a detailed comparison of independently generated GSMs as well as the construction of consensus GSMs that more comprehensively describe the knowledge on the modeled organism.GSMs can predict whether or not the corresponding organism and derived mutants can grow in a large variety of different growth conditions. In comparison, experimental data is extremely limited. For example, BIOLOG data describes growth phenotypes for one strain in a few hundred different media, and genome-wide gene essentially data is typically limited to a single growth medium. In chapter 4 GSMs of multiple Pseudomonas species were used to predict growth phenotypes for all possible single-gene-deletion mutants in all possible minimal growth media to determine conditionally and unconditionally essential genes. This simulated data was integrated with genomic data on 432 sequenced Pseudomonas species, which revealed a clear link between the essentiality of a gene function and the persistence of the gene within the Pseudomonas genus.Chapters 5 and 6 describe the use of GSMs to (re-)design microbial systems. P. putida is, despite its acknowledged versatile metabolism, an obligate aerobe. As the oxygen-requirement limits the potential applications of P. putida, there have been several experimental attempts to enable it to grow anaerobically, which have so far not succeeded. Chapter 5 describes an in silico effort to determine why P. putida cannot grow anaerobically using a combination of GSM analyses and comparative genomics. These analyses resulted in a shortlist of several essential and oxygen-dependent processes in P. putida. The identification of these processes has enabled the design of P. putida strains that can grow anaerobically based on the current understanding of P. putida metabolism as represented in GSMs.Efficient microbial CO2 fixation is a requirement for the biobased community, but the natural CO2 fixation pathways are rather inefficient, while the synthetic CO2 fixation pathways have been designed without considering the metabolic context of a target organism. Chapter 6 introduces a computational tool, CO2FIX, that designs species-specific CO2 fixation pathways based on GSMs and biochemical reaction databases. The designed pathways are evaluated for their ATP efficiency, thermodynamic feasibility, and kinetic rates. CO2FIX is applied to eight different organisms, which has led to the identification of both species-specific and general CO2 fixation pathways that have promising features while requiring surprisingly few non-native reactions. Three of these pathways are described in detail.In all previous chapters GSMs of relatively well-understood microbes have been used to gain further insight into their metabolism and to functionally (re-)design them. For complex microbial systems, such as algae (chapter 7) and gut microbial communities (chapter 8), GSMs are similarly useful, but substantially more difficult to create and analyze. Algae are widely considered as potential centerpieces of a biobased economy. Chapter 7 reviews the current challenges in algal genome annotation, modeling and synthetic biology. The gut microbiota is an incredibly complex microbial system that is crucial to our well-being. Chapter 8 reviews the ongoing developments in the modeling of both single gut microbes and gut microbial communities, and discusses how these developments will enable the move from studying correlation to causation, and ultimately the rational steering of gut microbial activity.Chapter 9 discusses how the previous chapters contribute to the research goals of this thesis. In addition, it provides an extensive discussion on current GSM practices, the issues associated therewith, and how these issues can be tackled. In particular, the discussion focuses on issues related to: (i) The inability to distinguish between biological difference and GSM generation artifacts when using multiple GSMs, (ii) The lack of continuous GSM updates, (iii) The mismatch between what GSM predictions and experimental data represent, (iv) The need for standardization in GSM evaluation, and (v) The lack of experimental validation of GSM-driven strain design for metabolic engineering.