Stochastic clonal expansion of "superstars" enhances the reserve capacity of enteric nervous system precursor cells.

We quantified cell population increase in the quail embryo enteric nervous system (ENS) from E2.5 (about 1500 cells) to E12 (about 8 million cells). We then probed ENS proliferative capacity by grafting to the chorio-allantoic membrane large (600 cells) and small (40 cells) populations of enteric neural crest (ENC) cells with aneural gut. This demonstrated that ENC cells show an extremely high capacity to regulate their proliferation while forming the ENS. Previous mathematical models and clonal label experiments revealed that a few dominant ENS "superstar" cell clones emerge but most clones are small. The model implied that "superstars" arise stochastically, but the same outcome could arise if "superstars" were pre-determined. We investigated these two modes mathematically and by grafting experiments with large and small numbers of ENCs, each including one EGFP-labelled ENC cell. The stochastic model predicts that the frequency of "superstar" detection increases as the ENC population decreases, the pre-determined model does not. Experimentally, as predicted by the stochastic model, the frequency of "superstar" detection increased with small ENC cell number. We conclude that ENS "superstar" clones achieve this status stochastically. Clonal dominance implies that clonal diversity is greatly reduced and in this case, somatic mutations may affect the phenotype. We suggest that somatic mutations coupled with loss of clonal diversity may contribute to variable penetrance and expressivity in individuals with genetically identical ENS pathologies.

Dynamic stepwise opening of integron attC DNA hairpins by SSB prevents toxicity and ensures functionality.

Biologically functional DNA hairpins are found in archaea, prokaryotes and eukaryotes, playing essential roles in various DNA transactions. However, during DNA replication, hairpin formation can stall the polymerase and is therefore prevented by the single-stranded DNA binding protein (SSB). Here, we address the question how hairpins maintain their functional secondary structure despite SSB's presence. As a model hairpin, we used the recombinogenic form of the attC site, essential for capturing antibiotic-resistance genes in the integrons of bacteria. We found that attC hairpins have a conserved high GC-content near their apical loop that creates a dynamic equilibrium between attC fully opened by SSB and a partially structured attC-6-SSB complex. This complex is recognized by the recombinase IntI, which extrudes the hairpin upon binding while displacing SSB. We anticipate that this intriguing regulation mechanism using a base pair distribution to balance hairpin structure formation and genetic stability is key to the dissemination of antibiotic resistance genes among bacteria and might be conserved among other functional hairpins.

Differential Clonal Expansion in an Invading Cell Population: Clonal Advantage or Dumb Luck?

In neoplastic cell growth, clones and subclones are variable both in size and mutational spectrum. The largest of these clones are believed to represent those cells with mutations that make them the most "fit," in a Darwinian sense, for expansion in their microenvironment. Thus, the degree of quantitative clonal expansion is regarded as being determined by innate qualitative differences between the cells that originate each clone. Here, using a combination of mathematical modelling and clonal labelling experiments applied to the developmental model system of the forming enteric nervous system, we describe how cells which are qualitatively identical may consistently produce clones of dramatically different sizes: most clones are very small while a few clones we term "superstars" contribute most of the cells to the final population. The basis of this is minor stochastic variations ("luck") in the timing and direction of movement and proliferation of individual cells, which builds a local advantage for daughter cells that is cumulative. This has potentially important consequences. In cancers, especially before strongly selective cytotoxic therapy, the assumption that the largest clones must be the cells with deterministic proliferative ability may not always hold true. In development, the gradual loss of clonal diversity as "superstars" take over the population may erode the resilience of the system to somatic mutations, which may have occurred early in clonal growth.

Spatial and temporal dynamics of cell generations within an invasion wave: a link to cell lineage tracing.

Mathematical models of a cell invasion wave have included both continuum partial differential equation (PDE) approaches and discrete agent-based cellular automata (CA) approaches. Here we are interested in modelling the spatial and temporal dynamics of the number of divisions (generation number) that cells have undergone by any time point within an invasion wave. In the CA framework this is performed from agent lineage tracings, while in the PDE approach a multi-species generalized Fisher equation is derived for the cell density within each generation. Both paradigms exhibit qualitatively similar cell generation densities that are spatially organized, with agents of low generation number rapidly attaining a steady state (with average generation number increasing linearly with distance) behind the moving wave and with evolving high generation number at the wavefront. This regularity in the generation spatial distributions is in contrast to the highly stochastic nature of the underlying lineage dynamics of the population. In addition, we construct a method for determining the lineage tracings of all agents without labelling and tracking the agents, but through either a knowledge of the spatial distribution of the generations or the number of agents in each generation. This involves determining generation-dependent proliferation probabilities and using these to define a generation-dependent Galton-Watson (GDGW) process. Monte-Carlo simulations of the GDGW process are used to determine the individual lineage tracings. The lineages of the GDGW process are analyzed using Lorenz curves and found to be similar to outcomes generated by direct lineage tracing in CA realizations. This analysis provides the basis for a potentially useful technique for deducing cell lineage data when imaging every cell is not feasible.

STENCIL-NET for equation-free forecasting from data.

We present an artificial neural network architecture, termed STENCIL-NET, for equation-free forecasting of spatiotemporal dynamics from data. STENCIL-NET works by learning a discrete propagator that is able to reproduce the spatiotemporal dynamics of the training data. This data-driven propagator can then be used to forecast or extrapolate dynamics without needing to know a governing equation. STENCIL-NET does not learn a governing equation, nor an approximation to the data themselves. It instead learns a discrete propagator that reproduces the data. It therefore generalizes well to different dynamics and different grid resolutions. By analogy with classic numerical methods, we show that the discrete forecasting operators learned by STENCIL-NET are numerically stable and accurate for data represented on regular Cartesian grids. A once-trained STENCIL-NET model can be used for equation-free forecasting on larger spatial domains and for longer times than it was trained for, as an autonomous predictor of chaotic dynamics, as a coarse-graining method, and as a data-adaptive de-noising method, as we illustrate in numerical experiments. In all tests, STENCIL-NET generalizes better and is computationally more efficient, both in training and inference, than neural network architectures based on local (CNN) or global (FNO) nonlinear convolutions.

Stability selection enables robust learning of differential equations from limited noisy data.

We present a statistical learning framework for robust identification of differential equations from noisy spatio-temporal data. We address two issues that have so far limited the application of such methods, namely their robustness against noise and the need for manual parameter tuning, by proposing stability-based model selection to determine the level of regularization required for reproducible inference. This avoids manual parameter tuning and improves robustness against noise in the data. Our stability selection approach, termed PDE-STRIDE, can be combined with any sparsity-promoting regression method and provides an interpretable criterion for model component importance. We show that the particular combination of stability selection with the iterative hard-thresholding algorithm from compressed sensing provides a fast and robust framework for equation inference that outperforms previous approaches with respect to accuracy, amount of data required, and robustness. We illustrate the performance of PDE-STRIDE on a range of simulated benchmark problems, and we demonstrate the applicability of PDE-STRIDE on real-world data by considering purely data-driven inference of the protein interaction network for embryonic polarization in Caenorhabditis elegans. Using fluorescence microscopy images of C. elegans zygotes as input data, PDE-STRIDE is able to learn the molecular interactions of the proteins.

Parallel Discrete Convolutions on Adaptive Particle Representations of Images.

We present data structures and algorithms for native implementations of discrete convolution operators over Adaptive Particle Representations (APR) of images on parallel computer architectures. The APR is a content-adaptive image representation that locally adapts the sampling resolution to the image signal. It has been developed as an alternative to pixel representations for large, sparse images as they typically occur in fluorescence microscopy. It has been shown to reduce the memory and runtime costs of storing, visualizing, and processing such images. This, however, requires that image processing natively operates on APRs, without intermediately reverting to pixels. Designing efficient and scalable APR-native image processing primitives, however, is complicated by the APR's irregular memory structure. Here, we provide the algorithmic building blocks required to efficiently and natively process APR images using a wide range of algorithms that can be formulated in terms of discrete convolutions. We show that APR convolution naturally leads to scale-adaptive algorithms that efficiently parallelize on multi-core CPU and GPU architectures. We quantify the speedups in comparison to pixel-based algorithms and convolutions on evenly sampled data. We achieve pixel-equivalent throughputs of up to 1TB/s on a single Nvidia GeForce RTX 2080 gaming GPU, requiring up to two orders of magnitude less memory than a pixel-based implementation.

Learning physically consistent differential equation models from data using group sparsity.

We propose a statistical learning framework based on group-sparse regression that can be used to (i) enforce conservation laws, (ii) ensure model equivalence, and (iii) guarantee symmetries when learning or inferring differential-equation models from data. Directly learning interpretable mathematical models from data has emerged as a valuable modeling approach. However, in areas such as biology, high noise levels, sensor-induced correlations, and strong intersystem variability can render data-driven models nonsensical or physically inconsistent without additional constraints on the model structure. Hence, it is important to leverage prior knowledge from physical principles to learn biologically plausible and physically consistent models rather than models that simply fit the data best. We present the group iterative hard thresholding algorithm and use stability selection to infer physically consistent models with minimal parameter tuning. We show several applications from systems biology that demonstrate the benefits of enforcing priors in data-driven modeling.

Adaptive particle representation of fluorescence microscopy images.

Modern microscopes create a data deluge with gigabytes of data generated each second, and terabytes per day. Storing and processing this data is a severe bottleneck, not fully alleviated by data compression. We argue that this is because images are processed as grids of pixels. To address this, we propose a content-adaptive representation of fluorescence microscopy images, the Adaptive Particle Representation (APR). The APR replaces pixels with particles positioned according to image content. The APR overcomes storage bottlenecks, as data compression does, but additionally overcomes memory and processing bottlenecks. Using noisy 3D images, we show that the APR adaptively represents the content of an image while maintaining image quality and that it enables orders of magnitude benefits across a range of image processing tasks. The APR provides a simple and efficient content-aware representation of fluosrescence microscopy images.

Cell lineage tracing in the developing enteric nervous system: superstars revealed by experiment and simulation.

Cell lineage tracing is a powerful tool for understanding how proliferation and differentiation of individual cells contribute to population behaviour. In the developing enteric nervous system (ENS), enteric neural crest (ENC) cells move and undergo massive population expansion by cell division within self-growing mesenchymal tissue. We show that single ENC cells labelled to follow clonality in the intestine reveal extraordinary and unpredictable variation in number and position of descendant cells, even though ENS development is highly predictable at the population level. We use an agent-based model to simulate ENC colonization and obtain agent lineage tracing data, which we analyse using econometric data analysis tools. In all realizations, a small proportion of identical initial agents accounts for a substantial proportion of the total final agent population. We term these individuals superstars. Their existence is consistent across individual realizations and is robust to changes in model parameters. This inequality of outcome is amplified at elevated proliferation rate. The experiments and model suggest that stochastic competition for resources is an important concept when understanding biological processes which feature high levels of cell proliferation. The results have implications for cell-fate processes in the ENS.