Posters

Poster presentations at ISMB 2020 will be presented virtually. Authors will pre-record their poster talk (5-7 minutes) and will upload it to the virtual conference platform site along with a PDF of their poster. All registered conference participants will have access to the poster and presentation through the conference and content until October 31, 2020. There are Q&A opportunities through a chat function to allow interaction between presenters and participants.

Preliminary information on preparing your poster and poster talk are available at: https://www.iscb.org/ismb2020-general/presenterinfo#posters

Ideally authors should be available for interactive chat during the times noted below:

Search:

View Posters By Category

Poster Session A: July 13 & July 14 7:45 am - 9:15 am Eastern Daylight Time	Session B: July 15 and July 16 between 7:45 am - 9:15 am Eastern Daylight Time
Bioinfo-core CAMDA COSI COVID-19 Education COSI EvoCompGen COSI Function / CAFA 4 MLCSB COSI SCANGEN (Special Session) SysMod COSI Systems Immunology (Special Session) Text Mining July 14 between 10:40 am - 2:00 pm EDT iRNA COSI	3DSIG COSI Bio-Ontologies COSI BioVis COSI CompMS COSI COVID-19 EvoCompGen COSI HitSeq COSI MICROBIOME COSI NetBio COSI RegSys COSI TransMed COSI VarI COSI General Comp Bio

Short Abstract: The bone marrow microenvironment is characterized by the existence of compartments with low oxygen partial pressure (a condition called hypoxia). The primary response of cells to hypoxia is the induction of hypoxia-inducible factors (HIF1 and HIF2), a transcription factor with many targets. This in turn triggers several changes, in particular in iron homeostasis and at the metabolic level. In this project we focus on Acute Myeloid Leukemia (AML), a set of genetically heterogeneous cancers of generally poor prognosis. We built a dynamical model of the network involved in the response to hypoxia of progenitor (human) cells. It comprises: the iron homeostasis module; signaling through the REDD1-mTOR pathway; energy homeostasis and the energy sensor AMPK; glucose uptake and ATP production.
Experiments specifically designed to build this model are underway. The model is expressed in terms of parameterized ordinary differential equations. We will present a simplified version of the model, with parameter values estimated from data obtained within the project. The goal is to provide an alternative approach to classify patients at diagnosis based on functional data with the aim of
eventually providing improved personalized therapeutic options.

Short Abstract: Biological rhythms are key features in eukaryotic cells. Coordinated oscillation in the expression of thousands of genes have been studied in multiple systems as ultradian or circadian rhythms, expression is reportedly coupled with changes in cell’s epigenetic landscape and metabolic oscillations. Although biological rhythms are approached in multiple studies, the complications of putting together different -omics information makes it difficult to gather integrative conclusions.
In this study, we present a novel analytical strategy for multi-omics analyses in biological rhythms. We extracted a multi-omics dataset from the Yeast Metabolic Cycle (YMC), a model system whose genes have been reported to cycle in three functional phases, coordinated with histone modifications and metabolomics changes.
We modelled this multi-omics dataset, which includes RNA-Seq, NET-Seq, H3K9ac and H3K18ac histone modifications, metabolomics and ATAC-Seq; and applied multivariate statistical methods to study their capacity to influence RNA-Seq. PLS and GLMs successfully modeled gene expression changes and showed that H3K18ac has a higher impact in gene expression, pointed the key metabolites that accumulate coordinating gene expression changes and extract Transcription Factors that drive gene expression. Finally, we applied PLS-pm for the first time in biological rhythms to create a global map of the regulatory orchestration in the YMC.

Short Abstract: DNA replication is a complex process that ensures genetic information maintenance. As recently observed, DNA replication timing is highly correlated with chromatin folding and global nuclear architecture. Here, we present a stochastic hybrid model of DNA replication that incorporates protein mobility and three-dimensional chromatin structure. Our model provides a framework to realistically simulate DNA replication for a complete eukaryotic genome and investigate the relationship between three-dimensional chromatin conformation and replication timing. Performing simulations for three model variants and a broad range of parameter values, we collected about 300,000 in silico replication profiles for fission yeast. We find that the number of firing factors initiating replication is rate-limiting and dominates the DNA replication completion time. We also find that explicitly modeling the recruitment of firing factors by the spindle pole body best reproduces experimental data and provide an independent validation of these findings in vivo. Further investigation of replication kinetics confirmed earlier observations of a rate-limiting number of firing factors in conjunction with their recycling upon replication fork collision. While the model faithfully reproduces global patterns of replication initiation, additional analysis of firing concurrence suggests that a uniform binding probability is too simplistic to capture local neighborhood effects in origin firing.

Short Abstract: Recently, the development of synthetic biology enables the construction of an artificial cell with genomic DNA designed from scratch. However, in contrast with rapid technological advances in DNA synthesis, the way to design genome is yet immature. A whole cell modeling is one of the powerful tools to design genome by predicting phenotype. However, it generally requires exclusive knowledge and advanced techniques. Here, we introduce a novel workflow for bottom-up modeling of a bacterial cell, Escherichia coli, from its genomic sequence. This workflow automates modeling, simulation, and visualization by integrating knowledge and experimental data on databases. First, it annotates genomic regions such as operons, ORFs, and regulatory domains according to the given sequence. Secondly, it simulates a genomic-scale model consisting of gene expression, protein modification, metabolism, and replication. The simulation results are quantitatively comparable with wet omics experiments. Finally, an interactive dashboard is deployed, where the “in silico omics” data is visualized by bioinformatics components such as genomic sequence, protein composition and pathway viewers. Due to the workflow, users can evaluate the phenotypic effects of sequence editing just by updating the input sequence. Our approach will provide a platform for genome design by integration of bioinformatics and systems biology.

Short Abstract: microRNAs are potent regulators of gene expression, and each may regulate dozens of genes across multiple pathways. Understanding the interplay of microRNA regulation of large scale molecular processes, coordinating the activity of multiple pathways, is still an open problem.

CARAWAY, (Capturing mirnA-controlled cooRdinated pAthWay ActivitY), identifies miRNAs that concurrently regulate functionally-related or coordinately expressed pathways. CARAWAY has the ability to prioritize miRNA orchestration of large-scale trans-pathway transcriptional programs via space and time efficient heuristics coupled with a miRNA:pathway-specific statistical model. These approaches enable CARAWAY to efficiently prioritize miRNAs interacting at small or large scale.

We have applied CARAWAY to discover miRNAs coordinating resilience against Alzheimer’s disease (AD) in individuals with pathological features of AD but no dementia. We have compared CARAWAY against standard and state-of-the-art approaches such as enrichment, or meta-analysis statistics. CARAWAY significantly outperforms the current methods, which exhibit clear bias towards prioritizing miRNAs with larger numbers of targeted genes. CARAWAY provides a sensitive and unbiased approach to detect miRNAs coordinating groups of biological pathways, providing a new level of insight into miRNA regulation of complex cellular processes.

Short Abstract: Cells show remarkable resilience against perturbations. Its evolutionary origin remains obscure. In order to leverage methods of systems biology to examine cellular robustness, a computationally accessible way of quantification is needed. Here, we present an unbiased metric of structural robustness in (genome-scale) metabolic models based on concepts prevalent in reliability engineering.

The probability of failure (PoF) is defined as the (weighted) portion of all combinations of loss-of-function mutations that disable network functionality. It can be exactly determined if all essential reactions, all synthetic lethal pairs of reactions, all synthetic lethal triplets of reactions etc., are known. In theory, these minimal cut sets (MCSs) can be calculated for any network, but for large models the problem remains computationally intractable. We demonstrate that even at the genome-scale only the lowest-cardinality MCSs are required to efficiently approximate the PoF.

We analyzed the robustness of 459 E. coli, Shigella, and Salmonella strains. In contrast to the congruence theory, which explains the origin of genetic robustness as a byproduct of selection for environmental flexibility, we found no correlation between robustness and the diversity of growth-supporting environments. On the contrary, our analysis indicates that the core-reactome, i.e. the set of reactions shared across strains, dominates robustness.

Short Abstract: Germinal centers (GCs) are specialized microanatomical structures that emerge within the secondary lymphoid organs upon infection or immunization and that play a central role in mounting an effective immune response. In this work, we present a stochastic multiscale model of the GC that includes an abstract representation of individual B cell receptors(BCRs) as strings of nucleotides. In addition, our model also accounts for the transcriptional program associated with B cell differentiation and the random cellular interactions that shape the GC reaction. The explicit sequence-based BCR representation allows us to track the effect of mutation son the BCR affinity to antigen by means of phylogenetic trees, which we compare to sequenced BCR repertoires. We extensively investigate the emergence of clonal dominance, which is observed in a subset of GCs. We find that small advantages in affinity acquired through random mutations are amplified and result in clonal dominance within a week, suggesting hence that stochastic effects play a fundamental role in the emergence of clonal bursts and dominance. Our faithful modelling of the GC reaction has broad implications in our understanding of the mechanisms underlying autoimmune disease and lymphomas, and it also contributes towards in silico vaccine design.

Short Abstract: Minimal organisms represent a stepping stone for the rational design of entire genomes. Mesoplasma florum, a fast-growing near-minimal organism for which genetic engineering techniques have been developed, is an interesting model for this task. Using sequence and structural homology, the set of metabolic functions encoded in its genome was identified, allowing the reconstruction of a metabolic network covering ~30% of its proteome. Experimental biomass composition, defined media compositions, substrate uptake and secretion rates were integrated as species-specific constraints to produce a functional model. Sensitivity analysis revealed oxygen dependency for the secretion of acetate, consistent with M. florum’s known facultative anaerobe phenotype. The model was validated and refined using genome-wide expression and essentiality datasets as well as growth data on varying carbohydrates. Discrepancies between model predictions and observations were mechanistically explained using protein structures and network analysis. The validated model, along with essentiality data and the complete transcription units architecture were used for the design of a reduced genome, thereby targeting 167 genes for removal.

Short Abstract: Increased demand for systems-level understanding of cellular metabolism has led to the widespread use of genome-scale metabolic models. However, analysis of large-scale systems, such as human metabolic networks, still presents considerable challenges for determining the systemic effects of individual metabolic perturbations. In this study, we have developed a model-based method for distinguishing between multiple metabolic conditions by highlighting the most significant differences in their flux spaces. Our approach first calculates the principal components describing each flux space. Subsequent network-based comparison of components between conditions yields the components which explain the differences between the flux spaces. The principal components also decompose the achievable steady-state flux states into low-dimensional reaction sets. The reaction sets thus obtained from all the conditions shed light on the metabolic features characterising each state. The biggest advantage of our approach is its scalability to even the largest models currently available (humans). As an application, we investigated the differences in the human adipocyte network arising due to presence/absence of metabolite uptake and identified the network-wide effects caused by the metabolic perturbation. We strongly believe that our method can be used to provide mechanistic insights into condition-specific metabolic changes in large genome-scale metabolic models.

Short Abstract: The immune system and its ability to make coordinated responses declines with age. Many efforts focus on deciphering underlying mechanisms in humans with the goal of improving immunity in the elderly. However, small cohorts, mixtures of cell types and unimodal profiling make it challenging to interpret the effects of aging. We performed an integrative multimodal and multi-variate analysis to couple age-associated changes in transcriptomic profiles with DNA methylation in 13 purified human immune cell types from 75 healthy individuals (ages 20-90 years old). Regression analyses showed similar effects with regard to numbers and extent of age-associated changes in most cell types. Only a minority of changes were shared across all cell types indicating that aging affects each cell type in a unique way. We leveraged age-dependent methylation and gene expression profiles within and across cell types to assess the effect of methylation on gene expression (mediation analysis). Identification of several such cell-specific loci and their relationships to disease-associated single nucleotide polymorphisms will be presented. Our observations provide insights into the relationship between transcriptomic and epigenetic basis of immune aging in humans.

Short Abstract: Protein folding, the determination of the lowest-energy configuration of a protein, is an unsolved computational problem. If protein folding could be solved, it would lead to significant advances in molecular biology, and technological development in areas such as drug discovery and catalyst design. As a hard combinatorial optimisation problem, protein folding has been studied as a potential target problem for adiabatic quantum computing. Although several experimental implementations have been discussed in the literature, the computational scaling of these approaches has not been elucidated. In this article, we present a numerical study of the (stoquastic) adiabatic quantum algorithm applied to protein lattice folding. Using exact numerical modelling of small systems, we find that the time-to-solution metric scales exponentially with peptide length, even for small peptides. However, comparison with classical heuristics for optimisation indicates a potential limited quantum speedup. Overall, our results suggest that quantum algorithms may well offer improvements for problems in the protein folding and structure prediction realm.

Short Abstract: Most cancer treatment induces DNA damage to eliminate highly proliferative cells. This process also affects healthy cells creating side effects. Cells respond differently to DNA damage depending on where they are in their cell cycle at the time of the treatment. Intriguingly, circadian rhythms control the timing of cell divisions. Personalized chronotherapy could be designed based on this observation: fitting the timing of the treatment to the patient’s natural daily rhythm. We built a mathematical model integrating the circadian rhythm, cell cycle, DNA damage response in Neurospora crassa and performed in-silico experiments. We simulated the perturbation of the circadian clock and the cell cycle upon DNA damage and analysed the consequences using phase response curves. We observed that DNA damage induces significant phase shifts during the subjective day in contrast to minimal phase shift during the subjective night. We analysed the behaviour of the cell cycle, where the cell cycle time and the strength of coupling between the circadian rhythm and cell cycle was noisy. We observed, depending on the timing of the induction of DNA damage, the healthy cells react differently: they can continue their normal division or they can be perturbed as the side effect of the treatment.

Short Abstract: The core circadian clock comprises a network of specialized neurons that regulate many essential rhythmic processes throughout an animal’s body. Its function, both endogenous and entrained, is paramount to ensuring health. Computational models of this clock often involve multiple ODEs per cell, and are expensive to simulate extensively. In this work, we introduce a method to reduce a high dimensional multicellular model of the circadian network to a lower dimensional phase-based model that preserves key features and enables efficient exploration of entrainment and control related problems. Our method preserves rhythmicity, cellular heterogeneity, and the ability to adjust phase via exogenous inputs. We use the reduced model in conjunction with the full model to explore two optimal control problems relevant: adjust the cell population phase in minimum time, and adjust the system’s phase using optimal effort. These problems inform the design of interventions for acute and chronic circadian misalignment (i.e., jetlag, seasonal affective disorders, or shiftwork). Preliminary results demonstrate that cellular heterogeneity enables greater ability to entrain and control the system, up to a limit. Insights gained from our findings can guide clinical interventions and enable better design of synthetic biological systems involving large populations of cells with cell-to-cell communication.

Short Abstract: Cell-cell adhesion can dictate tissue growth and multicellular pattern formation and it is crucial for the cellular dynamics during embryogenesis and cancer progression. While it is known that these adhesive forces are generated by cell adhesion molecules (CAMs), the regulation of CAMs is not well understood due to complex nonlinear interactions that span multiple levels of biological organization–from genetic regulation to whole-organism shape formation. We present a novel continuous model that can explain the dynamic relationships between genetic regulation, CAM expression, and differential adhesion. This approach can demonstrate the mechanisms responsible for cell-sorting behaviors, cell intercalation in proliferating populations, and the involution of zebrafish germ layer cells during gastrulation. The model can predict the physical parameters controlling the amplitude and wavelength of a cellular intercalation interface as shown in vitro. We demonstrate the crucial role of N-cadherin regulation for the involution and migration of cells beyond the gradient of the morphogen Nodal during zebrafish gastrulation. Integrating the emergent spatial tissue behaviors with the regulation of genes responsible for essential cellular properties such as adhesion will pave the way toward understanding the genetic regulation of large-scale complex patterns and shapes formation in developmental, regenerative, and cancer biology.

Short Abstract: Healthy wound healing exhibits an orderly succession of three phases: inflammation, proliferation, and resolution. In contrast, chronic wounds fail at proliferation, with competing hypotheses to explain why. Most computational models of wound healing have focused on either the inflammatory phase or the proliferative phase, but in this work, we develop a dynamical system model that studies the transitions between the three phases. Healthy healing is represented by a trajectory that dwells in three stable (or pseudo-stable) steady-states in sequence. The variables of our model represent key regulators of healthy and chronic wounds: bacteria, immune cells, cytokines, proteases, fibroblasts, and collagen matrix. We identify sensitive elements of the model that can, when slightly perturbed, plummet the entire system into a non-healing oscillation. Oscillation utilizes the capacity of a healthy wound to move in reverse, for example if re-injured and re-infected. Furthermore, we show that wound oscillation would appear static when assessed using the time-integrated variables that are clinically observable. In summary, we propose novel dynamical system explanation for why chronic wounds appear to have halted healing, and for integrating multiple previously-hypothesized culprits into a holistic model of disorganized information in the phase transitions of a self-organizing system.

Short Abstract: Plants’ innate immune response to pathogens is supported by a resilient signaling network that facilitates a robust and specific response when presented with a pathogen. The plant immune signaling network is large (hundreds of molecular components), and its dynamics are complex (e.g. transient response with multiple peaks). We sought a modeling framework based on ordinary differential equations (ODE) that was sufficiently flexible to be applied to a reduced network with complex dynamics. We found that the Multi-Compartment Modeling (MCM) framework, an ODE-based approach used frequently in other domains such as pharmacokinetics and epidemiology, is suitable for this purpose. Each compartment in the model operates as an ODE with first-order decay. A series of compartments, which can be considered a signaling pathway, were used to model single-peak dynamics, with a delayed peak time at each compartment. Output signals from two compartments in series were linearly combined as input signals to a third compartment, and this combination produced double-peak dynamics, consistent with our observations of some genes’ transcription following Arabidopsis exposure to a pathogen. We applied this MCM platform to fit genome-wide Arabidopsis transcriptional profiles in response to pathogen exposure and characterize the distinct dynamic mechanisms of different modes of plant immunity.

Short Abstract: Accurate and effective diagnosis, prognosis and treatment for cancer is difficult, in part because varying lifestyles among patients result in a wide range of responses to available treatments. Analyzing genomic signatures associated with important response factors could aid the selection of a correct diagnosis and the most effective treatment for individual patients, as opposed to a holistic assessment or a generalized treatment method. The goal of this study was to identify genomic signatures of smoking from gene expression/mutation data, and to build a predictor for smoking status of patients with bladder cancer. Gene data were taken from a 2014 Urothelial Bladder Carcinoma Study, mapped to a readable dataframe, and analyzed with the J48 package in R. This analysis produced a predictive classification tree, limited by a minimum frequency of occurrence in the data for each node (n=7). The model is compared with known pathways changes related to smoking. Studies are ongoing to increase the accuracy of the smoking status predictor in cancer patients by accounting for other habits that may interfere with the same signatures. This model could be used in conjunction with other predictors by physiologists to immediately suggest a likely-effective treatment to a patient without trial and error.

Short Abstract: Reverse engineering of molecular networks from biological data is one of the most challenging tasks in systems biology. Numerous inference and computational methodologies have been formalized to enhance the deduction of reliable and testable predictions in today’s biology. However, there is little research aimed to quantify how well the existing state-of-the-art molecular networks correspond to the measured gene expressions. We present a computational framework that combines formulation of probabilistic graphical model, standard statistical estimation, and integration of high-throughput gene expression data. The model is represented as a probabilistic bipartite graph, which accommodates partial information of diverse biological entities to study and analyze the global behavior of a biological system. This result provides a building block for performing simulations on the consistency between inferred gene regulatory networks and corresponding biological data. We test the applicability of our model to explore the allowable stable-states in two experimentally verified regulatory pathways in Escherechia Coli using real microarray expression data from the M3D database. Furthermore, the model is employed to quantify how well the pathways are explained by the extant microarray data. Results show a surprisingly high correlation between the observed states of the experimental data under various conditions and the inferred system’s behavior.

Short Abstract: Recent developments in mass spectrometry (MS) enable high-throughput screening of phospho-proteins across a broad range of biological contexts. Phospho-proteomic data complemented by computational algorithms enable the inference of kinase activity facilitating the identification of dysregulated kinases in various diseases, including cancer, Alzheimer’s disease and Parkinson’s disease, among others. However, the inadequacy of known kinase-substrate associations and the incompleteness of MS-based phosphorylation data pose important limitations on the inference of kinase activity. With a view to enhancing the reliability of kinase activity inference, we present a network-based framework named RoKAI that integrates various sources of functional information. These functional information include structure distance, co-evolution evidence, shared kinase associations, and protein-protein interaction networks. By propagating phosphorylation data across these networks, RoKAI obtains representative phosphorylation profiles capturing coordinated changes in signaling. The resulting phosphorylation profiles can be used in conjunction with any existing or future inference methods to predict kinase activity. The results of our computational experiments show that RoKAI consistently improves the accuracy of commonly used kinase activity inference methods and makes them more robust to missing kinase-substrate annotations. To provide an easy to use interface to users, RoKAI is available as a web-based tool at rokai.ngrok.io.

Short Abstract: The coronavirus disease 2019 (COVID-19) pandemic outbreak caused by the new coronavirus (SARS-CoV-2) is currently responsible for over 351 thousand deaths in 217 countries across the world (www.who.int/emergencies/diseases/novel-coronavirus-2019). The absence of FDA approved drugs against the new SARS-CoV-2 virus has prompted the urgent need to design new drugs and treatment management strategies against COVID-19.

We provide a combination of structural and dynamic modelling approaches to predict new drug targets against the SARS-CoV-2 virus and to determine drug optimisation strategies. This methodology involves the analysis of a stoichiometric metabolic model that integrates cell metabolism in humans with the SARS-CoV-2 virus. We also compared the interactions that occur between the Angiotensin-converting enzyme 2 (ACE2), the cellular receptor for the 2002 SARS-CoV virus, the new coronavirus (SARS-CoV-2) and associated cofactors. The model has provided an in-silico comparison of the biochemical demands of the viruses versus the host cells and predicted 18 essential reactions as drug targets from an integrated human cell and SARS-CoV-2 virus system. We are expanding the model to predict the effect of various treatment regimens to ensure maximum drug optimisation strategies against the virus.

Short Abstract: The activity of the adaptive immune system is largely governed by T cells and their specific T cell receptors (TCR), which selectively recognize foreign antigens. Recent advances in high-throughput sequencing techniques have enabled the sequencing of TCR repertoires and their antigenic targets, allowing us to research the missing link between TCR sequence and antigen binding specificity. Several machine learning approaches have achieved good results extrapolating to unseen TCRs, but no published study has managed to design and train an algorithm that is able to extrapolate to unseen epitopes. This severely limits the applicability of these algorithms, as the number of epitopes with known binding TCRs is small.
Here we propose a novel, multimodal neural network that allows to study independently the generalization capabilities to unseen TCRs and/or epitopes. Our algorithm utilizes an interpretable context attention mechanism that selectively highlights relevant oligopeptide and is designed to scale well to unseen epitopes. Achieving this would could advance a multitude of fields, including prediction of autoantigens in autoimmune diseases, development of immunotherapies for cancer, or informed vaccine design.

Short Abstract: Microbes naturally co-exist in communities, where they predominantly interact through the exchange of metabolites. Beyond natural communities, synthetic communities have been employed in the overproduction of biomolecules. Previous studies have shown that microbes in a community co-evolve to fulfill a desired functionality. Thus, it is interesting to understand the stability of interactions in a microbial community, as they co-evolve. In this work, we employ a novel computational approach to study the co-evolution of two-member microbial communities in silico, when subject to different community-level selection pressures. Specifically, we study pairwise communities of organisms, which exhibit different interaction types such as mutualism, amensalism and parasitism, and analyse the stability of the interactions upon co-evolution in silico. We predict the nature of interactions between organisms in a community metabolic network, using flux balance analysis. We identify the fitness benefits conferred upon evolution and investigate the metabolites exchanged among the organisms in the communities. Interestingly, when the selection pressure is altered to higher growth of organisms in the community, we observe a much higher tendency of the communities to evolve towards co-operation, i.e. mutualism. Overall, our study throws light on important aspects of microbial co-evolution and the stability of microbial interactions.

Short Abstract: The transition out of mitosis in budding yeast is controlled by the Mitotic Exit Network (MEN). The network interprets spatio-temporal cues about the progression of mitosis and ensures that release of Cdc14 phosphatase occurs only after completion of key mitotic events. The MEN has been studied intensively however a unified understanding of how localization and protein activity function together as a system is lacking. In this paper we present a compartmental, logical model of the MEN that can represent spatial aspects of regulation in parallel to control of enzymatic activity. Through optimization of the model, we reveal insights into role of Cdc5 in Cdc15 localization and the importance of Lte1 regulation in control of Bfa1. We show that our model is capable of correctly predicting the phenotype of roughly 80% of mutants we tested, including mutants representing mislocalizing proteins. We also demonstrate that the model can make predictions about the timing of mitotic exit and checkpoint competence in MEN mutants. Finally, we use the model to predict the impact of forced localization of MEN proteins and validate these predictions experimentally. This model represents a unified view of the mechanism of mitotic exit control.

Short Abstract: Chinese Hamster Ovary (CHO) cells belong to the most important organisms for the production of biopharmaceuticals. Metabolic modelling methods, such as flux balance analysis (FBA) have a potential to predict engineering strategies to increase protein production. To make these modelling tools available for CHO, a genome-scale metabolic model of CHO cells was built in an international collaboration. The composition of biomass is a key factor in the accuracy of FBA predictions. However the current model’s biomass is, due to the lack of available CHO-specific data, based on data from a mix of non-CHO sources. Here we close this gap by providing a comprehensive overview of the biomass composition in exponential phase of 13 different cell lines / conditions, including hosts and producers cell lines. We have observed big variability in cell dry mass, DNA, RNA and carbohydrate content and differences in lipid composition. Although the total protein content was significantly different, the amino acid composition was fairly stable for all cell lines and conditions. Uptake and secretion rates of amino acids, glucose, lactate and ammonium were determined. Finally, feeding these data into the metabolic model improved the growth rate predictions made by FBA and the reliability of the model.