Multi-signal regulation of the GSK-3ß homolog Rim11 controls meiosis entry in budding yeast.

Starvation in diploid budding yeast cells triggers a cell-fate program culminating in meiosis and spore formation. Transcriptional activation of early meiotic genes (EMGs) hinges on the master regulator Ime1, its DNA-binding partner Ume6, and GSK-3ß kinase Rim11. Phosphorylation of Ume6 by Rim11 is required for EMG activation. We report here that Rim11 functions as the central signal integrator for controlling Ume6 phosphorylation and EMG transcription. In nutrient-rich conditions, PKA suppresses Rim11 levels, while TORC1 retains Rim11 in the cytoplasm. Inhibition of PKA and TORC1 induces Rim11 expression and nuclear localization. Remarkably, nuclear Rim11 is required, but not sufficient, for Rim11-dependent Ume6 phosphorylation. In addition, Ime1 is an anchor protein enabling Ume6 phosphorylation by Rim11. Subsequently, Ume6-Ime1 coactivator complexes form and induce EMG transcription. Our results demonstrate how various signaling inputs (PKA/TORC1/Ime1) converge through Rim11 to regulate EMG expression and meiosis initiation. We posit that the signaling-regulatory network elucidated here generates robustness in cell-fate control.

Modelling capture efficiency of single-cell RNA-sequencing data improves inference of transcriptome-wide burst kinetics.

MOTIVATION: Gene expression is characterized by stochastic bursts of transcription that occur at brief and random periods of promoter activity. The kinetics of gene expression burstiness differs across the genome and is dependent on the promoter sequence, among other factors. Single-cell RNA sequencing (scRNA-seq) has made it possible to quantify the cell-to-cell variability in transcription at a global genome-wide level. However, scRNA-seq data are prone to technical variability, including low and variable capture efficiency of transcripts from individual cells. RESULTS: Here, we propose a novel mathematical theory for the observed variability in scRNA-seq data. Our method captures burst kinetics and variability in both the cell size and capture efficiency, which allows us to propose several likelihood-based and simulation-based methods for the inference of burst kinetics from scRNA-seq data. Using both synthetic and real data, we show that the simulation-based methods provide an accurate, robust and flexible tool for inferring burst kinetics from scRNA-seq data. In particular, in a supervised manner, a simulation-based inference method based on neural networks proves to be accurate and useful when applied to both allele and nonallele-specific scRNA-seq data. AVAILABILITY AND IMPLEMENTATION: The code for Neural Network and Approximate Bayesian Computation inference is available at https://github.com/WT215/nnRNA and https://github.com/WT215/Julia_ABC, respectively.

Efficient Bayesian inference for stochastic agent-based models.

The modelling of many real-world problems relies on computationally heavy simulations of randomly interacting individuals or agents. However, the values of the parameters that underlie the interactions between agents are typically poorly known, and hence they need to be inferred from macroscopic observations of the system. Since statistical inference rests on repeated simulations to sample the parameter space, the high computational expense of these simulations can become a stumbling block. In this paper, we compare two ways to mitigate this issue in a Bayesian setting through the use of machine learning methods: One approach is to construct lightweight surrogate models to substitute the simulations used in inference. Alternatively, one might altogether circumvent the need for Bayesian sampling schemes and directly estimate the posterior distribution. We focus on stochastic simulations that track autonomous agents and present two case studies: tumour growths and the spread of infectious diseases. We demonstrate that good accuracy in inference can be achieved with a relatively small number of simulations, making our machine learning approaches orders of magnitude faster than classical simulation-based methods that rely on sampling the parameter space. However, we find that while some methods generally produce more robust results than others, no algorithm offers a one-size-fits-all solution when attempting to infer model parameters from observations. Instead, one must choose the inference technique with the specific real-world application in mind. The stochastic nature of the considered real-world phenomena poses an additional challenge that can become insurmountable for some approaches. Overall, we find machine learning approaches that create direct inference machines to be promising for real-world applications. We present our findings as general guidelines for modelling practitioners.

Benchmarking imputation methods for network inference using a novel method of synthetic scRNA-seq data generation.

BACKGROUND: Single cell RNA-sequencing (scRNA-seq) has very rapidly become the new workhorse of modern biology providing an unprecedented global view on cellular diversity and heterogeneity. In particular, the structure of gene-gene expression correlation contains information on the underlying gene regulatory networks. However, interpretation of scRNA-seq data is challenging due to specific experimental error and biases that are unique to this kind of data including drop-out (or technical zeros). METHODS: To deal with this problem several methods for imputation of zeros for scRNA-seq have been developed. However, it is not clear how these processing steps affect inference of genetic networks from single cell data. Here, we introduce Biomodelling.jl, a tool for generation of synthetic scRNA-seq data using multiscale modelling of stochastic gene regulatory networks in growing and dividing cells. RESULTS: Our tool produces realistic transcription data with a known ground truth network topology that can be used to benchmark different approaches for gene regulatory network inference. Using this tool we investigate the impact of different imputation methods on the performance of several network inference algorithms. CONCLUSIONS: Biomodelling.jl provides a versatile and useful tool for future development and benchmarking of network inference approaches using scRNA-seq data.

Integrating multi-OMICS data through sparse canonical correlation analysis for the prediction of complex traits: a comparison study.

MOTIVATION: Recent developments in technology have enabled researchers to collect multiple OMICS datasets for the same individuals. The conventional approach for understanding the relationships between the collected datasets and the complex trait of interest would be through the analysis of each OMIC dataset separately from the rest, or to test for associations between the OMICS datasets. In this work we show that integrating multiple OMICS datasets together, instead of analysing them separately, improves our understanding of their in-between relationships as well as the predictive accuracy for the tested trait. Several approaches have been proposed for the integration of heterogeneous and high-dimensional (p≫n) data, such as OMICS. The sparse variant of canonical correlation analysis (CCA) approach is a promising one that seeks to penalize the canonical variables for producing sparse latent variables while achieving maximal correlation between the datasets. Over the last years, a number of approaches for implementing sparse CCA (sCCA) have been proposed, where they differ on their objective functions, iterative algorithm for obtaining the sparse latent variables and make different assumptions about the original datasets. RESULTS: Through a comparative study we have explored the performance of the conventional CCA proposed by Parkhomenko et al., penalized matrix decomposition CCA proposed by Witten and Tibshirani and its extension proposed by Suo et al. The aforementioned methods were modified to allow for different penalty functions. Although sCCA is an unsupervised learning approach for understanding of the in-between relationships, we have twisted the problem as a supervised learning one and investigated how the computed latent variables can be used for predicting complex traits. The approaches were extended to allow for multiple (more than two) datasets where the trait was included as one of the input datasets. Both ways have shown improvement over conventional predictive models that include one or multiple datasets. AVAILABILITY AND IMPLEMENTATION: https://github.com/theorod93/sCCA. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

bayNorm: Bayesian gene expression recovery, imputation and normalization for single-cell RNA-sequencing data.

MOTIVATION: Normalization of single-cell RNA-sequencing (scRNA-seq) data is a prerequisite to their interpretation. The marked technical variability, high amounts of missing observations and batch effect typical of scRNA-seq datasets make this task particularly challenging. There is a need for an efficient and unified approach for normalization, imputation and batch effect correction. RESULTS: Here, we introduce bayNorm, a novel Bayesian approach for scaling and inference of scRNA-seq counts. The method's likelihood function follows a binomial model of mRNA capture, while priors are estimated from expression values across cells using an empirical Bayes approach. We first validate our assumptions by showing this model can reproduce different statistics observed in real scRNA-seq data. We demonstrate using publicly available scRNA-seq datasets and simulated expression data that bayNorm allows robust imputation of missing values generating realistic transcript distributions that match single molecule fluorescence in situ hybridization measurements. Moreover, by using priors informed by dataset structures, bayNorm improves accuracy and sensitivity of differential expression analysis and reduces batch effect compared with other existing methods. Altogether, bayNorm provides an efficient, integrated solution for global scaling normalization, imputation and true count recovery of gene expression measurements from scRNA-seq data. AVAILABILITY AND IMPLEMENTATION: The R package 'bayNorm' is publishd on bioconductor at https://bioconductor.org/packages/release/bioc/html/bayNorm.html. The code for analyzing data in this article is available at https://github.com/WT215/bayNorm_papercode. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The impact of random microenvironmental fluctuations on tumor control probability.

The tumor control probability (TCP) is a metric used to calculate the probability of controlling or eradicating tumors through radiotherapy. Cancer cells vary in their response to radiation, and although many factors are involved, the tumor microenvironment is a crucial one that determines radiation efficacy. The tumor microenvironment plays a significant role in cancer initiation and propagation, as well as in treatment outcome. We have developed stochastic formulations to study the impact of arbitrary microenvironmental fluctuations on TCP and extinction probability (EP), which is defined as the probability of cancer cells removal in the absence of treatment. Since the derivation of analytical solutions are not possible for complicated cases, we employ a modified Gillespie algorithm to analyze TCP and EP, considering the random variations in cellular proliferation and death rates. Our results show that increasing the standard deviation in kinetic rates initially enhances the probability of tumor eradication. However, if the EP does not reach a probability of 1, the increase in the standard deviation subsequently has a negative impact on probability of cancer cells removal, decreasing the EP over time. The greatest effect on EP has been observed when both birth and death rates are being randomly modified and are anticorrelated. In addition, similar results are observed for TCP, where radiotherapy is included, indicating that increasing the standard deviation in kinetic rates at first enhances the probability of tumor eradication. But, it has a negative impact on treatment effectiveness if the TCP does not reach a probability of 1.

A bacterial size law revealed by a coarse-grained model of cell physiology.

Universal observations in Biology are sometimes described as "laws". In E. coli, experimental studies performed over the past six decades have revealed major growth laws relating ribosomal mass fraction and cell size to the growth rate. Because they formalize complex emerging principles in biology, growth laws have been instrumental in shaping our understanding of bacterial physiology. Here, we discovered a novel size law that connects cell size to the inverse of the metabolic proteome mass fraction and the active fraction of ribosomes. We used a simple whole-cell coarse-grained model of cell physiology that combines the proteome allocation theory and the structural model of cell division. This integrated model captures all available experimental data connecting the cell proteome composition, ribosome activity, division size and growth rate in response to nutrient quality, antibiotic treatment and increased protein burden. Finally, a stochastic extension of the model explains non-trivial correlations observed in single cell experiments including the adder principle. This work provides a simple and robust theoretical framework for studying the fundamental principles of cell size determination in unicellular organisms.

The influence of nuclear compartmentalisation on stochastic dynamics of self-repressing gene expression.

Gene expression is an inherently noisy process. This noise is generally thought to be deleterious as precise internal regulation of biochemical reactions is essential for cell growth and survival. Self-repression of gene expression, which is the simplest form of a negative feedback loop, is commonly believed to be employed by cellular systems to decrease the stochastic fluctuations in gene expression. When there is some delay in autoregulation, it is also believed that this system can generate oscillations. In eukaryotic cells, mRNAs that are synthesised in the nucleus must be exported to the cytoplasm to function in protein synthesis, whereas proteins must be transported into the nucleus from the cytoplasm to regulate the expression levels of genes. Nuclear transport thus plays a critical role in eukaryotic gene expression and regulation. Some recent studies have suggested that nuclear retention of mRNAs can control noise in mRNA expression. However, the effect of nuclear transport on protein noise and its interplay with negative feedback regulation is not completely understood. In this paper, we systematically compare four different simple models of gene expression. By using simulations and applying the linear noise approximation to the corresponding chemical master equations, we investigate the influence of nuclear import and export on noise in gene expression in a negative autoregulatory feedback loop. We first present results consistent with the literature, i.e., that negative feedback can effectively buffer the variability in protein levels, and nuclear retention can decrease mRNA noise levels. Interestingly we find that when negative feedback is combined with nuclear retention, an amplification in gene expression noise can be observed and is dependant on nuclear translocation rates. Finally, we investigate the effect of nuclear compartmentalisation on the ability of self-repressing genes to exhibit stochastic oscillatory dynamics.

The scaffold protein Ste5 directly controls a switch-like mating decision in yeast.

Evolution has resulted in numerous innovations that allow organisms to increase their fitness by choosing particular mating partners, including secondary sexual characteristics, behavioural patterns, chemical attractants and corresponding sensory mechanisms. The haploid yeast Saccharomyces cerevisiae selects mating partners by interpreting the concentration gradient of pheromone secreted by potential mates through a network of mitogen-activated protein kinase (MAPK) signalling proteins. The mating decision in yeast is an all-or-none, or switch-like, response that allows cells to filter weak pheromone signals, thus avoiding inappropriate commitment to mating by responding only at or above critical concentrations when a mate is sufficiently close. The molecular mechanisms that govern the switch-like mating decision are poorly understood. Here we show that the switching mechanism arises from competition between the MAPK Fus3 and a phosphatase Ptc1 for control of the phosphorylation state of four sites on the scaffold protein Ste5. This competition results in a switch-like dissociation of Fus3 from Ste5 that is necessary to generate the switch-like mating response. Thus, the decision to mate is made at an early stage in the pheromone pathway and occurs rapidly, perhaps to prevent the loss of the potential mate to competitors. We argue that the architecture of the Fus3-Ste5-Ptc1 circuit generates a novel ultrasensitivity mechanism, which is robust to variations in the concentrations of these proteins. This robustness helps assure that mating can occur despite stochastic or genetic variation between individuals. The role of Ste5 as a direct modulator of a cell-fate decision expands the functional repertoire of scaffold proteins beyond providing specificity and efficiency of information processing. Similar mechanisms may govern cellular decisions in higher organisms and be disrupted in cancer.

Biosensor architectures for high-fidelity reporting of cellular signaling.

Understanding mechanisms of information processing in cellular signaling networks requires quantitative measurements of protein activities in living cells. Biosensors are molecular probes that have been developed to directly track the activity of specific signaling proteins and their use is revolutionizing our understanding of signal transduction. The use of biosensors relies on the assumption that their activity is linearly proportional to the activity of the signaling protein they have been engineered to track. We use mechanistic mathematical models of common biosensor architectures (single-chain FRET-based biosensors), which include both intramolecular and intermolecular reactions, to study the validity of the linearity assumption. As a result of the classic mechanism of zero-order ultrasensitivity, we find that biosensor activity can be highly nonlinear so that small changes in signaling protein activity can give rise to large changes in biosensor activity and vice versa. This nonlinearity is abolished in architectures that favor the formation of biosensor oligomers, but oligomeric biosensors produce complicated FRET states. Based on this finding, we show that high-fidelity reporting is possible when a single-chain intermolecular biosensor is used that cannot undergo intramolecular reactions and is restricted to forming dimers. We provide phase diagrams that compare various trade-offs, including observer effects, which further highlight the utility of biosensor architectures that favor intermolecular over intramolecular binding. We discuss challenges in calibrating and constructing biosensors and highlight the utility of mathematical models in designing novel probes for cellular signaling.

Multi-view data visualisation via manifold learning.

Non-linear dimensionality reduction can be performed by manifold learning approaches, such as stochastic neighbour embedding (SNE), locally linear embedding (LLE) and isometric feature mapping (ISOMAP). These methods aim to produce two or three latent embeddings, primarily to visualise the data in intelligible representations. This manuscript proposes extensions of Student's t-distributed SNE (t-SNE), LLE and ISOMAP, for dimensionality reduction and visualisation of multi-view data. Multi-view data refers to multiple types of data generated from the same samples. The proposed multi-view approaches provide more comprehensible projections of the samples compared to the ones obtained by visualising each data-view separately. Commonly, visualisation is used for identifying underlying patterns within the samples. By incorporating the obtained low-dimensional embeddings from the multi-view manifold approaches into the K-means clustering algorithm, it is shown that clusters of the samples are accurately identified. Through extensive comparisons of novel and existing multi-view manifold learning algorithms on real and synthetic data, the proposed multi-view extension of t-SNE, named multi-SNE, is found to have the best performance, quantified both qualitatively and quantitatively by assessing the clusterings obtained. The applicability of multi-SNE is illustrated by its implementation in the newly developed and challenging multi-omics single-cell data. The aim is to visualise and identify cell heterogeneity and cell types in biological tissues relevant to health and disease. In this application, multi-SNE provides an improved performance over single-view manifold learning approaches and a promising solution for unified clustering of multi-omics single-cell data.

Global transcription regulation revealed from dynamical correlations in time-resolved single-cell RNA sequencing.

Single-cell transcriptomics reveals significant variations in transcriptional activity across cells. Yet, it remains challenging to identify mechanisms of transcription dynamics from static snapshots. It is thus still unknown what drives global transcription dynamics in single cells. We present a stochastic model of gene expression with cell size- and cell cycle-dependent rates in growing and dividing cells that harnesses temporal dimensions of single-cell RNA sequencing through metabolic labeling protocols and cel lcycle reporters. We develop a parallel and highly scalable approximate Bayesian computation method that corrects for technical variation and accurately quantifies absolute burst frequency, burst size, and degradation rate along the cell cycle at a transcriptome-wide scale. Using Bayesian model selection, we reveal scaling between transcription rates and cell size and unveil waves of gene regulation across the cell cycle-dependent transcriptome. Our study shows that stochastic modeling of dynamical correlations identifies global mechanisms of transcription regulation. A record of this paper's transparent peer review process is included in the supplemental information.

A coarse-grained resource allocation model of carbon and nitrogen metabolism in unicellular microbes.

Coarse-grained resource allocation models (C-GRAMs) are simple mathematical models of cell physiology, where large components of the macromolecular composition are abstracted into single entities. The dynamics and steady-state behaviour of such models provides insights on optimal allocation of cellular resources and have explained experimentally observed cellular growth laws, but current models do not account for the uptake of compound sources of carbon and nitrogen. Here, we formulate a C-GRAM with nitrogen and carbon pathways converging on biomass production, with parametrizations accounting for respirofermentative and purely respiratory growth. The model describes the effects of the uptake of sugars, ammonium and/or compound nutrients such as amino acids on the translational resource allocation towards proteome sectors that maximized the growth rate. It robustly recovers cellular growth laws including the Monod law and the ribosomal growth law. Furthermore, we show how the growth-maximizing balance between carbon uptake, recycling, and excretion depends on the nutrient environment. Lastly, we find a robust linear correlation between the ribosome fraction and the abundance of amino acid equivalents in the optimal cell, which supports the view that simple regulation of translational gene expression can enable cells to achieve an approximately optimal growth state.

Fission yeast obeys a linear size law under nutrient titration.

Steady-state cell size and geometry depend on growth conditions. Here, we use an experimental setup based on continuous culture and single-cell imaging to study how cell volume, length, width and surface-to-volume ratio vary across a range of growth conditions including nitrogen and carbon titration, the choice of nitrogen source, and translation inhibition. Overall, we find cell geometry is not fully determined by growth rate and depends on the specific mode of growth rate modulation. However, under nitrogen and carbon titrations, we observe that the cell volume and the growth rate follow the same linear scaling.

Selective advantage of epigenetically disrupted cancer cells via phenotypic inertia.

The evolution of established cancers is driven by selection of cells with enhanced fitness. Subclonal mutations in numerous epigenetic regulator genes are common across cancer types, yet their functional impact has been unclear. Here, we show that disruption of the epigenetic regulatory network increases the tolerance of cancer cells to unfavorable environments experienced within growing tumors by promoting the emergence of stress-resistant subpopulations. Disruption of epigenetic control does not promote selection of genetically defined subclones or favor a phenotypic switch in response to environmental changes. Instead, it prevents cells from mounting an efficient stress response via modulation of global transcriptional activity. This "transcriptional numbness" lowers the probability of cell death at early stages, increasing the chance of long-term adaptation at the population level. Our findings provide a mechanistic explanation for the widespread selection of subclonal epigenetic-related mutations in cancer and uncover phenotypic inertia as a cellular trait that drives subclone expansion.

Data-driven spatio-temporal modelling of glioblastoma.

Mathematical oncology provides unique and invaluable insights into tumour growth on both the microscopic and macroscopic levels. This review presents state-of-the-art modelling techniques and focuses on their role in understanding glioblastoma, a malignant form of brain cancer. For each approach, we summarize the scope, drawbacks and assets. We highlight the potential clinical applications of each modelling technique and discuss the connections between the mathematical models and the molecular and imaging data used to inform them. By doing so, we aim to prime cancer researchers with current and emerging computational tools for understanding tumour progression. By providing an in-depth picture of the different modelling techniques, we also aim to assist researchers who seek to build and develop their own models and the associated inference frameworks. Our article thus strikes a unique balance. On the one hand, we provide a comprehensive overview of the available modelling techniques and their applications, including key mathematical expressions. On the other hand, the content is accessible to mathematicians and biomedical scientists alike to accommodate the interdisciplinary nature of cancer research.

Growth-rate-dependent and nutrient-specific gene expression resource allocation in fission yeast.

Cellular resources are limited and their relative allocation to gene expression programmes determines physiological states and global properties such as the growth rate. Here, we determined the importance of the growth rate in explaining relative changes in protein and mRNA levels in the simple eukaryote Schizosaccharomyces pombe grown on non-limiting nitrogen sources. Although expression of half of fission yeast genes was significantly correlated with the growth rate, this came alongside wide-spread nutrient-specific regulation. Proteome and transcriptome often showed coordinated regulation but with notable exceptions, such as metabolic enzymes. Genes positively correlated with growth rate participated in every level of protein production apart from RNA polymerase II-dependent transcription. Negatively correlated genes belonged mainly to the environmental stress response programme. Critically, metabolic enzymes, which represent ∼55-70% of the proteome by mass, showed mostly condition-specific regulation. In summary, we provide a rich account of resource allocation to gene expression in a simple eukaryote, advancing our basic understanding of the interplay between growth-rate-dependent and nutrient-specific gene expression.

Emergent expression of fitness-conferring genes by phenotypic selection.

Genotypic and phenotypic adaptation is the consequence of ongoing natural selection in populations and is key to predicting and preventing drug resistance. Whereas classic antibiotic persistence is all-or-nothing, here we demonstrate that an antibiotic resistance gene displays linear dose-responsive selection for increased expression in proportion to rising antibiotic concentration in growing Escherichia coli populations. Furthermore, we report the potentially wide-spread nature of this form of emergent gene expression (EGE) by instantaneous phenotypic selection process under bactericidal and bacteriostatic antibiotic treatment, as well as an amino acid synthesis pathway enzyme under a range of auxotrophic conditions. We propose an analogy to Ohm's law in electricity (V = IR), where selection pressure acts similarly to voltage (V), gene expression to current (I), and resistance (R) to cellular machinery constraints and costs. Lastly, mathematical modeling using agent-based models of stochastic gene expression in growing populations and Bayesian model selection reveal that the EGE mechanism requires variability in gene expression within an isogenic population, and a cellular "memory" from positive feedbacks between growth and expression of any fitness-conferring gene. Finally, we discuss the connection of the observed phenomenon to a previously described general fluctuation-response relationship in biology.

Discrimination of Escherichia coli, Shigella flexneri, and Shigella sonnei using lipid profiling by MALDI-TOF mass spectrometry paired with machine learning.

Matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) has become a staple in clinical microbiology laboratories. Protein-profiling of bacteria using this technique has accelerated the identification of pathogens in diagnostic workflows. Recently, lipid profiling has emerged as a way to complement bacterial identification where protein-based methods fail to provide accurate results. This study aimed to address the challenge of rapid discrimination between Escherichia coli and Shigella spp. using MALDI-TOF MS in the negative ion mode for lipid profiling coupled with machine learning. Both E. coli and Shigella species are closely related; they share high sequence homology, reported for 16S rRNA gene sequence similarities between E. coli and Shigella spp. exceeding 99%, and a similar protein expression pattern but are epidemiologically distinct. A bacterial collection of 45 E. coli, 48 Shigella flexneri, and 62 Shigella sonnei clinical isolates were submitted to lipid profiling in negative ion mode using the MALDI Biotyper Sirius® system after treatment with mild-acid hydrolysis (acetic acid 1% v/v for 15 min at 98°C). Spectra were then analyzed using our in-house machine learning algorithm and top-ranked features used for the discrimination of the bacterial species. Here, as a proof-of-concept, we showed that lipid profiling might have the potential to differentiate E. coli from Shigella species using the analysis of the top five ranked features obtained by MALDI-TOF MS in the negative ion mode of the MALDI Biotyper Sirius® system. Based on this new approach, MALDI-TOF MS analysis of lipids might help pave the way toward these goals.