PLOS Computational Biology

Norovirus, primarily transmitted via fomite route, poses a significant threat to global public health and the economy. Airports, as critical transportation hubs connecting people from around the world, has high potential risk of norovirus transmission due to large number of public surfaces. A total of 21.3 hours of video episodes were recorded across nine functional areas at the airport, capturing 25,925 touches. A surface transmission model based on a Markov chain was developed. Using the beta-Poisson dose-response model, the infection risk of norovirus and the effectiveness of various interventions in different airports’ areas were quantified. Without any preventive measures, restaurants at airports exhibited the highest risk of norovirus transmission, with an infection probability of 8.8×10⁻³% (95% CI, 1.5×10⁻³% -2.1×10⁻²%). This means approximately 4.6 (95% CI, 0.8–10.9) out of 51,494 passengers who entered the restaurants would be infected by an infected passenger. Comparing with no surface disinfection, disinfecting public surfaces every 2 hours can reduce the risk of norovirus infection per visit to the airport by 83.2%. In contrast, comparing with no hand washing, handwashing every 2 hours can reduce the infection risk per visit to the airport by only 2.0%, making public surface disinfection significantly more effective than handwashing. If the mask-wearing rate increases from 0% to 50%, the infection risk of norovirus would be decreased by 48.0% (95% CI, 43.5–52.3%). Furthermore, using antimicrobial copper/copper-nickel alloy coatings for most public surfaces could reduce the infection risk by 15.9%-99.2%.

Households are a major driver of transmission of acute respiratory viruses, such as SARS-CoV-2 or Influenza. Until now antiviral treatments have mostly been used as a curative treatment in symptomatic individuals. During an outbreak, more aggressive strategies involving pre- or post-exposure prophylaxis (PrEP or PEP) could be employed to further reduce the risk of severe disease but also prevent transmission to household contacts. In order to understand the effectiveness of such strategies and the factors that may modulate them, we developed a multi-scale model that follows the infection at both the individual-level (viral dynamics) and the population-level (transmission dynamics) in households. Using a simulation study we explored different antiviral treatment strategies, evaluating their effectiveness on reducing the transmission risk and the virological burden in households for a range of virus characteristics (e.g., secondary attack rate—SAR, or time to peak viral load). We found that when the index case can be identified and treated before symptom onset, both transmission and virological burden are reduced by > 75% for most SAR values and time to peak viral load, with minimal benefit to treat additionally household contacts. While treatment initiated after index symptom onset does not reduce the risk of transmission, it can still reduce the virological burden in the household, a proxy for severe disease and subsequent transmission risk outside the household. In that case optimal strategies involve treatment of both index case and household contacts as PEP, with efficacy > 50% when peak viral load occurs after symptom onset, and 30-50% otherwise. In all the considered cases, antiviral treatment strategies were optimal for SAR ranging 20-60%, and for larger household sizes. This study highlights the opportunity of antiviral drug-based interventions in households during an outbreak to minimize viral transmission and disease burden.

Diversity plays an important role in various domains, including conservation, whether it describes diversity within a population or diversity over a set of species. While various strategies for measuring among-species diversity have emerged (e.g. Phylogenetic Diversity (PD), Split System Diversity (SSD) and entropy-based methods), extensions to populations are rare. An understudied problem is how to assess the diversity of a collection of populations where each has its own internal diversity. Relying solely on measures that treat each population as a monomorphic lineage (like a species) can be misleading. To address this problem, we present four population-level diversity assessment approaches: Pooling, Averaging, Pairwise Differencing, and Fixing. These approaches can be used to extend any diversity measure that is primarily defined for a group of individuals to a collection of populations. We then apply the approaches to two measures of diversity that have been used in conservation—Heterozygosity (Het) and Split System Diversity (SSD)—across a dataset comprising SNP data for 50 anadromous Atlantic salmon populations. We investigate agreement and disagreement between these measures of diversity when used to identify optimal sets of populations for conservation, on both the observed data, and randomized and simulated datasets. The similarity and differences of the maximum-diversity sets as well as the pairwise correlations among our proposed measures emphasize the need to clearly define what aspects of biodiversity we aim to both measure and optimize, to ensure meaningful and effective conservation decisions.

Proprioception is one of the least understood senses, yet fundamental for the control of movement. Even basic questions of how limb pose is represented in the somatosensory cortex are unclear. We developed a topographic variational autoencoder with lateral connectivity (topo-VAE) to compute a putative cortical map from a large set of natural movement data. Although not fitted to neural data, our model reproduces two sets of observations from monkey centre-out reaching: 1. The shape and velocity dependence of proprioceptive receptive fields in hand-centered coordinates despite the model having no knowledge of arm kinematics or hand coordinate systems. 2. The distribution of neuronal preferred directions (PDs) recorded from multi-electrode arrays. The model makes several testable predictions: 1. Encoding across the cortex has a blob-and-pinwheel-type geometry of PDs. 2. Few neurons will encode just a single joint. Our model provides a principled basis for understanding of sensorimotor representations, and the theoretical basis of neural manifolds, with applications to the restoration of sensory feedback in brain-computer interfaces and the control of humanoid robots.

Motivation DNA sequencing of multiple bulk samples from a tumor provides the opportunity to investigate tumor heterogeneity and reconstruct a phylogeny of a patient’s cancer. However, since bulk DNA sequencing of tumor tissue measures thousands of cells from a heterogeneous mixture of distinct sub-populations, accurate reconstruction of the tumor phylogeny requires simultaneous deconvolution of cancer clones and inference of ancestral relationships, leading to a challenging computational problem. Many existing methods for phylogenetic reconstruction from bulk sequencing data do not scale to large datasets, such as recent datasets containing upwards of ninety samples with dozens of distinct sub-populations. Results We develop an approach to reconstruct phylogenetic trees from multi-sample bulk DNA sequencing data by separating the reconstruction problem into two parts: a structured regression problem for a fixed tree T , and an optimization over tree space. We derive an algorithm for the regression sub-problem by exploiting the unique, combinatorial structure of the matrices appearing within the problem. This algorithm has both asymptotic and empirical improvements over linear programming (LP) approaches to the problem. Using our algorithm for this regression sub-problem, we develop fastBE , a simple method for phylogenetic inference from multi-sample bulk DNA sequencing data. We demonstrate on simulated data with hundreds of samples and upwards of a thousand distinct sub-populations that fastBE outperforms existing approaches in terms of reconstruction accuracy, sample efficiency, and runtime. Owing to its scalability, fastBE enables both phylogenetic reconstruction directly from indvidual mutations without requiring the clustering of mutations into clones, as well as a new phylogeny constrained mutation clustering algorithm. On real data from fourteen B-progenitor acute lymphoblastic leukemia patients, fastBE infers mutation phylogenies with fewer violations of a widely used evolutionary constraint and better agreement to the observed mutational frequencies. Using our phylogeny constrained mutation clustering algorithm, we also find mutation clusters with lower distortion compared to state-of-the-art approaches. Finally, we show that on two patient-derived colorectal cancer models, fastBE infers mutation phylogenies with less violation of a widely used evolutionary constraint compared to existing methods.

Neuroplasticity refers to functional and structural changes in brain regions in response to healthy and pathological activity. Activity dependent plasticity induced by epileptic activity can involve healthy brain regions into the epileptogenic network by perturbing their excitation/inhibition balance. In this article, we present a new neural mass model, which accounts for neuroplasticity, for investigating the possible mechanisms underlying the epileptogenic network expansion. Our multiple-timescale model is inspired by physiological calcium-mediated synaptic plasticity and pathological extrasynaptic N-methyl-D-aspartate (NMDA) dependent plasticity dynamics. The model highlights that synaptic plasticity at excitatory connections and structural changes in the inhibitory system can transform a healthy region into a secondary epileptic focus under recurrent seizures and interictal activity occurring in the primary focus. Our results suggest that the latent period of this transformation can provide a window of opportunity to prevent the expansion of epileptogenic networks, formation of an epileptic focus, or other comorbidities associated with epileptic activity.

Protein-RNA interactions play a critical role in many cellular processes and pathologies. However, experimental determination of protein-RNA structures is still challenging, therefore computational tools are needed for the prediction of protein-RNA interfaces. Although evolutionary pressures can be exploited for structural prediction of protein-protein interfaces, and recent deep learning methods using protein multiple sequence alignments have radically improved the performance of protein-protein interface structural prediction, protein-RNA structural prediction is lagging behind, due to the scarcity of structural data and the flexibility involved in these complexes. To study the evolution of protein-RNA interface structures, we first identified a large and diverse dataset of 2,022 pairs of structurally homologous interfaces (termed structural interologs). We leveraged this unique dataset to analyze the conservation of interface contacts among structural interologs based on the properties of involved amino acids and nucleotides. We uncovered that 73% of distance-based contacts and 68% of apolar contacts are conserved on average, and the strong conservation of these contacts occurs even in distant homologs with sequence identity below 20%. Distance-based contacts are also much more conserved compared to what we had found in a previous study of homologous protein-protein interfaces. In contrast, hydrogen bonds, salt bridges, and π-stacking interactions are very versatile in pairs of protein-RNA interologs, even for close homologs with high interface sequence identity. We found that almost half of the non-conserved distance-based contacts are linked to a small proportion of interface residues that no longer make interface contacts in the interolog, a phenomenon we term “interface switching out”. We also examined possible recovery mechanisms for non-conserved hydrogen bonds and salt bridges, uncovering diverse scenarios of switching out, change in amino acid chemical nature, intermolecular and intramolecular compensations. Our findings provide insights for integrating evolutionary signals into predictive protein-RNA structural modeling methods.

Gene expression patterns in developing organisms are established by groups of cross-regulating target genes that are driven by morphogen gradients. As development progresses, morphogen activity is reduced, leaving the emergent pattern without stabilizing positional cues and at risk of rapid deterioration due to the inherently noisy biochemical processes at the cellular level. But remarkably, gene expression patterns remain spatially stable and reproducible over long developmental time spans in many biological systems. Here we combine spatial-stochastic simulations with an enhanced sampling method (Non-Stationary Forward Flux Sampling) and a recently developed stability theory to address how spatiotemporal integrity of a gene expression pattern is maintained in developing tissue lacking morphogen gradients. Using a minimal embryo model consisting of spatially coupled biochemical reactor volumes, we study a prototypical stripe pattern in which weak cross-repression between nearest neighbor expression domains alternates with strong repression between next-nearest neighbor domains, inspired by the gap gene system in the Drosophila embryo. We find that tuning of the weak repressive interactions to an optimal level can prolong stability of the expression patterns by orders of magnitude, enabling stable patterns over developmentally relevant times in the absence of morphogen gradients. The optimal parameter regime found in simulations of the embryo model closely agrees with the predictions of our coarse-grained stability theory. To elucidate the origin of stability, we analyze a reduced phase space defined by two measures of pattern asymmetry. We find that in the optimal regime, intact patterns are protected via restoring forces that counteract random perturbations and give rise to a metastable basin. Together, our results demonstrate that metastable attractors can emerge as a property of stochastic gene expression patterns even without system-wide positional cues, provided that the gene regulatory interactions shaping the pattern are optimally tuned.

Theta-band oscillations (3–8 Hz) in the mammalian hippocampus organize the temporal structure of cortical inputs, resulting in a phase code that enables rhythmic input sampling for episodic memory formation and spatial navigation. However, it remains unclear what evolutionary pressures might have driven the selection of theta over higher-frequency bands that could potentially provide increased input sampling resolution. Here, we address this question by introducing a theoretical framework that combines the efficient coding and neural oscillatory sampling hypotheses, focusing on the information rate (bits/s) of phase coding neurons. We demonstrate that physiologically realistic noise levels create a trade-off between the speed of input sampling, determined by oscillation frequency, and encoding precision in rodent hippocampal neurons. This speed-precision trade-off results in a maximum information rate of ∼1–2 bits/s within the theta frequency band, thus confining the optimal oscillation frequency to the low end of the spectrum. We also show that this framework accounts for key hippocampal features, such as the preservation of the theta band along the dorsoventral axis despite physiological gradients, and the modulation of theta frequency and amplitude by running speed. Extending the analysis beyond the hippocampus, we propose that theta oscillations could also support efficient stimulus encoding in the visual cortex and olfactory bulb. More broadly, our framework lays the foundation for studying how system features, such as noise, constrain the optimal sampling frequencies in both biological and artificial brains.

Biological signalling systems are complex, and efforts to build mechanistic models must confront a huge parameter space, indirect and sparse data, and frequently encounter multiscale and multiphysics phenomena. We present HOSS, a framework for Hierarchical Optimization of Systems Simulations, to address such problems. HOSS operates by breaking down extensive systems models into individual pathway blocks organized in a nested hierarchy. At the first level, dependencies are solely on signalling inputs, and subsequent levels rely only on the preceding ones. We demonstrate that each independent pathway in every level can be efficiently optimized. Once optimized, its parameters are held constant while the pathway serves as input for succeeding levels. We develop an algorithmic approach to identify the necessary nested hierarchies for the application of HOSS in any given biochemical network. Furthermore, we devise two parallelizable variants that generate numerous model instances using stochastic scrambling of parameters during initial and intermediate stages of optimization. Our results indicate that these variants produce superior models and offer an estimate of solution degeneracy. Additionally, we showcase the effectiveness of the optimization methods for both abstracted, event-based simulations and ODE-based models.

When making risky choices in social contexts, humans typically combine social information with individual preferences about the options at stake. It remains unknown how such decisions are made when these preferences are inaccessible or disrupted, as might be the case for individuals confronting novel options or experiencing cognitive impairment. Thus, we examined participants with lesions in insular or dorsal anterior cingulate cortex, key regions implicated in risky decision-making, as they played a gambling task where choices were made both alone and after observing others’ choices. Participants in both lesion groups showed disrupted use of standard utility-based computations about risky options. For socially situated decisions, these participants showed increased conformity with the choices of others, independent from social utility-based computations. These findings suggest that in social contexts, following others’ choices may be a heuristic for decision-making when utility-based risk processing is disrupted.

Simulations of large-scale brain dynamics are often impacted by overexcitation resulting from heavy-tailed structural network distributions, leading to biologically implausible simulation results. We implement a homeodynamic plasticity mechanism, known from other modeling work, in the widely used Jansen-Rit neural mass model for The Virtual Brain (TVB) simulation framework. We aim at heterogeneously adjusting the inhibitory coupling weights to reach desired dynamic regimes in each brain region. We show that, by using this dynamic approach, we can control the target activity level to obtain biologically plausible brain simulations, including post-synaptic potentials and blood-oxygen-level-dependent functional magnetic resonance imaging (fMRI) activity. We demonstrate that the derived dynamic Feedback Inhibitory Control (dFIC) can be used to enable increased variability of model dynamics. We derive the conditions under which the simulated brain activity converges to a predefined target level analytically and via simulations. We highlight the benefits of dFIC in the context of fitting the TVB model to static and dynamic measures of fMRI empirical data, accounting for global synchronization across the whole brain. The proposed novel method helps computational neuroscientists, especially TVB users, to easily “tune” brain models to desired dynamical regimes depending on the specific requirements of each study. The presented method is a steppingstone towards increased biological realism in brain network models and a valuable tool to better understand their underlying behavior.

Microsatellite instability (MSI) is an important genomic biomarker for cancer diagnosis and treatment, and sequencing-based approaches are often applied to identify MSI because of its fastness and efficiency. These approaches, however, may fail to identify MSI on one or more sub-clones for certain cancers with a high degree of heterogeneity, leading to erroneous diagnoses and unsuitable treatments. Besides, the computational cost of identifying sub-clonal MSI can be exponentially increased when multiple sub-clones with different length distributions share MSI status. Herein, this paper proposes “scMSI”, an accurate and efficient estimation of sub-clonal MSI to identify the microsatellite status. scMSI is an integrative Bayesian method to deconvolute the mixed-length distribution of sub-clones by a novel alternating iterative optimization procedure based on a subtle generative model. During the process of deconvolution, the optimized division of each sub-clone is attained by a heuristic algorithm, aligning with clone proportions that adhere optimally to the sample’s clonal structure. To evaluate the performance, 16 patients diagnosed with endometrial cancer, exhibiting positive responses to the treatment despite having negative MSI status based on sequencing-based approaches, were considered. Excitingly, scMSI reported MSI on sub-clones successfully, and the findings matched the conclusions on immunohistochemistry. In addition, testing results on a series of experiments with simulation datasets concerning a variety of impact factors demonstrated the effectiveness and superiority of scMSI in detecting MSI on sub-clones over existing approaches. scMSI provides a new way of detecting MSI for cancers with a high degree of heterogeneity.

During morphogenesis, cells precisely generate forces that drive cell shape changes and cellular motion. These forces predominantly arise from contractility of the actomyosin cortex, allowing for cortical tension, protrusion formation, and cell division. Image-based force inference can derive such forces from microscopy images, without complicated and time-consuming experimental set-ups. However, current methods do not account for common effects, such as physical confinement and local force generation. Here we propose a force-inference method based on a biophysical model of cell shape, and assess relative cellular surface tension, adhesive tension between cells, as well as cytokinesis and protrusion formation. We applied our method on fluorescent microscopy images of the early C. elegans embryo. Predictions for cell surface tension at the 7-cell stage were validated by measurements using cortical laser ablation. Our non-invasive method facilitates the accurate tracking of force generation, and offers many new perspectives for studying morphogenesis.

Human facial shape, while strongly heritable, involves both genetic and structural complexity, necessitating precise phenotyping for accurate assessment. Common phenotyping strategies include simplifying 3D facial features into univariate traits such as anthropometric measurements (e.g., inter-landmark distances), unsupervised dimensionality reductions (e.g., principal component analysis (PCA) and auto-encoder (AE) approaches), and assessing resemblance to particular facial gestalts (e.g., syndromic facial archetypes). This study provides a comparative assessment of these strategies in genome-wide association studies (GWASs) of 3D facial shape. Specifically, we investigated inter-landmark distances, PCA and AE-derived latent dimensions, and facial resemblance to random, extreme, and syndromic gestalts within a GWAS of 8,426 individuals of recent European ancestry. Inter-landmark distances exhibit the highest SNP-based heritability as estimated via LD score regression, followed by AE dimensions. Conversely, resemblance scores to extreme and syndromic facial gestalts display the lowest heritability, in line with expectations. Notably, the aggregation of multiple GWASs on facial resemblance to random gestalts reveals the highest number of independent genetic loci. This novel, easy-to-implement phenotyping approach holds significant promise for capturing genetically relevant morphological traits derived from complex biomedical imaging datasets, and its applications extend beyond faces. Nevertheless, these different phenotyping strategies capture different genetic influences on craniofacial shape. Thus, it remains valuable to explore these strategies individually and in combination to gain a more comprehensive understanding of the genetic factors underlying craniofacial shape and related traits.

How well does the information contained in vocal signals travel through the environment? To assess the efficiency of information transfer in little auk ( Alle alle , an Arctic seabird) calls over distance, we selected two of the social call types with the highest potential for individuality coding. Using available recordings of known individuals, we calculated the apparent source levels, with apparent maximum peak sound pressure level (ASPL) of 63 dB re 20 μPa at 1 m for both call types. Further, we created a sound attenuation model using meteorological data collected in the vicinity of the little auk colony in Hornsund, Spitsbergen. Using this model, we modelled the calls to reflect higher frequency filtering and sound level loss occurring during spherical spreading in perfect local conditions, down to the putative hearing threshold of the species, calculated to equal ASPL of signals “propagated” to roughly one kilometre. Those modelled calls were then used in a permuted discriminant function analysis, support vector machine models, and linear models of Beecher’s information statistic, to investigate whether transmission loss will affect the retention of individual information of the signal. Calls could be correctly classified to individuals above chance level independently of the distance, down to and over the putative physiological hearing threshold. Interestingly, the information capacity of the signal did not decrease with its filtering and attenuation. While this study touches on signal properties purely and cannot provide evidence of the actual use by the animals, it shows that little auk signals can theoretically travel long distances with negligible information loss, and supports the hypothesis that vocalisations could facilitate long-distance communication in the species.

How do newborns learn to see? We propose that visual systems are space-time fitters, meaning visual development can be understood as a blind fitting process (akin to evolution) in which visual systems gradually adapt to the spatiotemporal data distributions in the newborn’s environment. To test whether space-time fitting is a viable theory for learning how to see, we performed parallel controlled-rearing experiments on newborn chicks and deep neural networks (DNNs), including CNNs and transformers. First, we raised newborn chicks in impoverished environments containing a single object, then simulated those environments in a video game engine. Second, we recorded first-person images from agents moving through the virtual animal chambers and used those images to train DNNs. Third, we compared the viewpoint-invariant object recognition performance of the chicks and DNNs. When DNNs received the same visual diet (training data) as chicks, the models developed common object recognition skills as chicks. DNNs that used time as a teaching signal—space-time fitters—also showed common patterns of successes and failures across the test viewpoints as chicks. Thus, DNNs can learn object recognition in the same impoverished environments as newborn animals. We argue that space-time fitters can serve as formal scientific models of newborn visual systems, providing image-computable models for studying how newborns learn to see from raw visual experiences.

Despite almost a century of research on energetics in biological systems, we still cannot explain energy regulation in social groups, like ant colonies. How do individuals regulate their collective activity without a centralized control system? What is the role of social interactions in distributing the workload amongst group members? And how does the group save energy by avoiding being constantly active? We offer new insight into these questions by studying an intuitive compartmental model, calibrated with and compared to data on ant colonies. The model describes a previously unexplored balance between positive and negative social feedback driven by individual activity: when activity levels are low, the presence of active individuals stimulates inactive individuals to start working; when activity levels are high, however, active individuals inhibit each other, effectively capping the proportion of active individuals at any one time. Through the analysis of the system stability, we demonstrate that this balance results in energetic spending at the group level growing proportionally slower than the group size. Our finding is reminiscent of Kleiber’s law of metabolic scaling in unitary organisms and highlights the critical role of social interactions in driving the collective energetic efficiency of group-living organisms.

Tuberculosis ( TB ), the disease caused by Mycobacterium tuberculosis ( Mtb ), remains a major health problem with 10.6 million cases of the disease and 1.6 million deaths in 2021. It is well understood that pulmonary TB is due to Mtb growth in the lung but quantitative estimates of rates of Mtb replication and death in lungs of patients or animals such as monkeys or rabbits remain largely unknown. We performed experiments with rabbits infected with a novel, virulent clinical Mtb isolate of the Beijing lineage, HN878, carrying an unstable plasmid pBP10. In our in vitro experiments we found that pBP10 is more stable in HN878 strain than in a more commonly used laboratory-adapted Mtb strain H37Rv (the segregation coefficient being s = 0.10 in HN878 vs. s = 0.18 in H37Rv). Interestingly, the kinetics of plasmid-bearing bacteria in lungs of Mtb-infected rabbits did not follow an expected monotonic decline; the percent of plasmid-bearing cells increased between 28 and 56 days post-infection and remained stable between 84 and 112 days post-infection despite a large increase in bacterial numbers in the lung at late time points. Mathematical modeling suggested that such a non-monotonic change in the percent of plasmid-bearing cells can be explained if the lung Mtb population consists of several (at least 2) sub-populations with different replication/death kinetics: one major population expanding early and being controlled/eliminated, while another, a smaller population expanding at later times causing a counterintuitive increase in the percent of plasmid-bearing cells. Importantly, a model with one kinetically homogeneous Mtb population could not explain the data including when the model was run stochastically. Given that in rabbits HN878 strain forms well circumscribed granulomas, our results suggest independent bacterial dynamics in subsets of such granulomas. Our model predictions can be tested in future experiments in which HN878-pBP10 dynamics in individual granulomas is followed over time. Taken together, our new data and mathematical modeling-based analyses illustrate differences in Mtb dynamics in mice and rabbits confirming a perhaps somewhat obvious observation that “rabbits are not mice”.

Research on cell differentiation facilitates a deeper understanding of the fundamental processes of life, elucidates the intrinsic mechanisms underlying diseases such as cancer, and advances the development of therapeutics and precision medicine. Existing methods for inferring cell differentiation trajectories from single-cell RNA sequencing (scRNA-seq) data primarily rely on static gene expression data to measure distances between cells and subsequently infer pseudotime trajectories. In this work, we introduce a novel method, scGRN-Entropy, for inferring cell differentiation trajectories and pseudotime from scRNA-seq data. Unlike existing approaches, scGRN-Entropy improves inference accuracy by incorporating dynamic changes in gene regulatory networks (GRN). In scGRN-Entropy, an undirected graph representing state transitions between cells is constructed by integrating both static relationships in gene expression space and dynamic relationships in the GRN space. The edges of the undirected graph are then refined using pseudotime inferred based on cell entropy in the GRN space. Finally, the Minimum Spanning Tree (MST) algorithm is applied to derive the cell differentiation trajectory. We validate the accuracy of scGRN-Entropy on eight different real scRNA-seq datasets, demonstrating its superior performance in inferring cell differentiation trajectories through comparative analysis with existing state-of-the-art methods.

Identifying goal-relevant features in novel environments is a central challenge for efficient behaviour. We asked whether humans address this challenge by relying on prior knowledge about common properties of reward-predicting features. One such property is the rate of change of features, given that behaviourally relevant processes tend to change on a slower timescale than noise. Hence, we asked whether humans are biased to learn more when task-relevant features are slow rather than fast. To test this idea, 295 human participants were asked to learn the rewards of two-dimensional bandits when either a slowly or quickly changing feature of the bandit predicted reward. Across two experiments and one preregistered replication, participants accrued more reward when a bandit’s relevant feature changed slowly, and its irrelevant feature quickly, as compared to the opposite. We did not find a difference in the ability to generalise to unseen feature values between conditions. Testing how feature speed could affect learning with a set of four function approximation Kalman filter models revealed that participants had a higher learning rate for the slow feature, and adjusted their learning to both the relevance and the speed of feature changes. The larger the improvement in participants’ performance for slow compared to fast bandits, the more strongly they adjusted their learning rates. These results provide evidence that human reinforcement learning favours slower features, suggesting a bias in how humans approach reward learning.

Many evolving ecosystems have spatial structures that can be conceptualized as networks, with nodes representing individuals or homogeneous subpopulations and links the patterns of spread between them. Prior models of evolution on networks do not take ecological niche differences and eco-evolutionary interplay into account. Here, we combine a resource competition model with evolutionary graph theory to study how heterogeneous topological structure shapes evolutionary dynamics under global frequency-dependent ecological interactions. We find that the addition of ecological competition for resources can produce a reversal of roles between amplifier and suppressor networks for deleterious mutants entering the population. We show that this effect is a nonlinear function of ecological niche overlap and discuss intuition for the observed dynamics using simulations and analytical approximations. We use these theoretical results together with spatial representations from imaging data to show that, for ductal carcinoma, where tumor growth is highly spatially constrained, with cells confined to a tree-like network of ducts, the topological structure can lead to higher rates of deleterious mutant hitchhiking with metabolic driver mutations, compared to tumors characterized by different spatial topologies.

Coexistence of multiple strains of a pathogen in a host population can present significant challenges to vaccine development or treatment efficacy. Here we discuss a novel mechanism that can increase rates of long-lived strain polymorphism, rooted in the presence of social structure in a host population. We show that social preference of interaction, in conjunction with differences in immunity between host subgroups, can exert varying selection pressure on pathogen strains, creating a balancing mechanism that supports stable viral coexistence, independent of other known mechanisms. We use population genetic models to study rates of pathogen heterozygosity as a function of population size, host population composition, mutant strain fitness differences and host social preferences of interaction. We also show that even small periodic epochs of host population stratification can lead to elevated strain coexistence. These results are robust to varying social preferences of interaction, overall differences in strain fitnesses, and spatial heterogeneity in host population composition. Our results highlight the role of host population social stratification in increasing rates of pathogen strain diversity, with effects that should be considered when designing policies or treatments with a long-term view of curbing pathogen evolution.

Measles is an important infectious disease system both for its burden on public health and as an opportunity for studying nonlinear spatio-temporal disease dynamics. Traditional mechanistic models often struggle to fully capture the complex nonlinear spatio-temporal dynamics inherent in measles outbreaks. In this paper, we first develop a high-dimensional feed-forward neural network model with spatial features (SFNN) to forecast endemic measles outbreaks and systematically compare its predictive power with that of a classical mechanistic model (TSIR). We illustrate the utility of our model using England and Wales measles data from 1944-1965. These data present multiple modeling challenges due to the interplay between metapopulations, seasonal trends, and nonlinear dynamics related to demographic changes. Our results show that while the TSIR model yields similarly performant short-term (1 to 2 biweeks ahead) forecasts for highly populous cities, our neural network model (SFNN) consistently achieves lower root mean squared error (RMSE) across other forecasting windows. Furthermore, we show that our spatial-feature neural network model, without imposing mechanistic assumptions a priori , can uncover gravity-model-like spatial hierarchy of measles spread in which major cities play an important role in driving regional outbreaks. We then turn our attention to integrative approaches that combine mechanistic and machine learning models. Specifically, we investigate how the TSIR can be utilized to improve a state-of-the-art approach known as Physics-Informed-Neural-Networks (PINN) which explicitly combines compartmental models and neural networks. Our results show that the TSIR can facilitate the reconstruction of latent susceptible dynamics, thereby enhancing both forecasts in terms of mean absolute error (MAE) and parameter inference of measles dynamics within the PINN. In summary, our results show that appropriately designed neural network-based models can outperform traditional mechanistic models for short to long-term forecasts, while simultaneously providing mechanistic interpretability. Our work also provides valuable insights into more effectively integrating machine learning models with mechanistic models to enhance public health responses to measles and similar infectious disease systems.