Ten simple rules for writing Dockerfiles for reproducible data science.

Computational science has been greatly improved by the use of containers for packaging software and data dependencies. In a scholarly context, the main drivers for using these containers are transparency and support of reproducibility; in turn, a workflow's reproducibility can be greatly affected by the choices that are made with respect to building containers. In many cases, the build process for the container's image is created from instructions provided in a Dockerfile format. In support of this approach, we present a set of rules to help researchers write understandable Dockerfiles for typical data science workflows. By following the rules in this article, researchers can create containers suitable for sharing with fellow scientists, for including in scholarly communication such as education or scientific papers, and for effective and sustainable personal workflows.

DeepClean: Self-Supervised Artefact Rejection for Intensive Care Waveform Data Using Deep Generative Learning.

Waveform physiological data are important in the treatment of critically ill patients in the intensive care unit. Such recordings are susceptible to artefacts, which must be removed before the data can be reused for alerting or reprocessed for other clinical or research purposes. Accurate removal of artefacts reduces bias and uncertainty in clinical assessment, as well as the false positive rate of ICU alarms, and is therefore a key component in providing optimal clinical care. In this work, we present DeepClean, a prototype self-supervised artefact detection system using a convolutional variational autoencoder deep neural network that avoids costly and painstaking manual annotation, requiring only easily obtained 'good' data for training. For a test case with invasive arterial blood pressure, we demonstrate that our algorithm can detect the presence of an artefact within a 10s sample of data with sensitivity and specificity around 90%. Furthermore, DeepClean was able to identify regions of artefacts within such samples with high accuracy, and we show that it significantly outperforms a baseline principal component analysis approach in both signal reconstruction and artefact detection. DeepClean learns a generative model and therefore may also be used for imputation of missing data.

Causality indices for bivariate time series data: A comparative review of performance.

Inferring nonlinear and asymmetric causal relationships between multivariate longitudinal data is a challenging task with wide-ranging application areas including clinical medicine, mathematical biology, economics, and environmental research. A number of methods for inferring causal relationships within complex dynamic and stochastic systems have been proposed, but there is not a unified consistent definition of causality in the context of time series data. We evaluate the performance of ten prominent causality indices for bivariate time series across four simulated model systems that have different coupling schemes and characteristics. Pairwise correlations between different methods, averaged across all simulations, show that there is generally strong agreement between methods, with minimum, median, and maximum Pearson correlations between any pair (excluding two similarity indices) of 0.298, 0.719, and 0.955, respectively. In further experiments, we show that these methods are not always invariant to real-world relevant transformations (data availability, standardization and scaling, rounding errors, missing data, and noisy data). We recommend transfer entropy and nonlinear Granger causality as particularly strong approaches for estimating bivariate causal relationships in real-world applications. Both successfully identify causal relationships and a lack thereof across multiple simulations, while remaining robust to rounding errors, at least 20% missing data and small variance Gaussian noise. Finally, we provide flexible open-access Python code for computation of these methods and for the model simulations.

meaRtools: An R package for the analysis of neuronal networks recorded on microelectrode arrays.

Here we present an open-source R package 'meaRtools' that provides a platform for analyzing neuronal networks recorded on Microelectrode Arrays (MEAs). Cultured neuronal networks monitored with MEAs are now being widely used to characterize in vitro models of neurological disorders and to evaluate pharmaceutical compounds. meaRtools provides core algorithms for MEA spike train analysis, feature extraction, statistical analysis and plotting of multiple MEA recordings with multiple genotypes and treatments. meaRtools functionality covers novel solutions for spike train analysis, including algorithms to assess electrode cross-correlation using the spike train tiling coefficient (STTC), mutual information, synchronized bursts and entropy within cultured wells. Also integrated is a solution to account for bursts variability originating from mixed-cell neuronal cultures. The package provides a statistical platform built specifically for MEA data that can combine multiple MEA recordings and compare extracted features between different genetic models or treatments. We demonstrate the utilization of meaRtools to successfully identify epilepsy-like phenotypes in neuronal networks from Celf4 knockout mice. The package is freely available under the GPL license (GPL> = 3) and is updated frequently on the CRAN web-server repository. The package, along with full documentation can be downloaded from: https://cran.r-project.org/web/packages/meaRtools/.

Homeostatic Activity-Dependent Tuning of Recurrent Networks for Robust Propagation of Activity.

Developing neuronal networks display spontaneous bursts of action potentials that are necessary for circuit organization and tuning. While spontaneous activity has been shown to instruct map formation in sensory circuits, it is unknown whether it plays a role in the organization of motor networks that produce rhythmic output. Using computational modeling, we investigate how recurrent networks of excitatory and inhibitory neuronal populations assemble to produce robust patterns of unidirectional and precisely timed propagating activity during organism locomotion. One example is provided by the motor network inDrosophilalarvae, which generates propagating peristaltic waves of muscle contractions during crawling. We examine two activity-dependent models, which tune weak network connectivity based on spontaneous activity patterns: a Hebbian model, where coincident activity in neighboring populations strengthens connections between them; and a homeostatic model, where connections are homeostatically regulated to maintain a constant level of excitatory activity based on spontaneous input. The homeostatic model successfully tunes network connectivity to generate robust activity patterns with appropriate timing relationships between neighboring populations. These timing relationships can be modulated by the properties of spontaneous activity, suggesting its instructive role for generating functional variability in network output. In contrast, the Hebbian model fails to produce the tight timing relationships between neighboring populations required for unidirectional activity propagation, even when additional assumptions are imposed to constrain synaptic growth. These results argue that homeostatic mechanisms are more likely than Hebbian mechanisms to tune weak connectivity based on spontaneous input in a recurrent network for rhythm generation and robust activity propagation. SIGNIFICANCE STATEMENT: How are neural circuits organized and tuned to maintain stable function and produce robust output? This task is especially difficult during development, when circuit properties change in response to variable environments and internal states. Many developing circuits exhibit spontaneous activity, but its role in the synaptic organization of motor networks that produce rhythmic output is unknown. We studied a model motor network, that when appropriately tuned, generates propagating activity as during crawling inDrosophilalarvae. Based on experimental evidence of activity-dependent tuning of connectivity, we examined plausible mechanisms by which appropriate connectivity emerges. Our results suggest that activity-dependent homeostatic mechanisms are better suited than Hebbian mechanisms for organizing motor network connectivity, and highlight an important difference from sensory areas.

A comparison of computational methods for detecting bursts in neuronal spike trains and their application to human stem cell-derived neuronal networks.

Accurate identification of bursting activity is an essential element in the characterization of neuronal network activity. Despite this, no one technique for identifying bursts in spike trains has been widely adopted. Instead, many methods have been developed for the analysis of bursting activity, often on an ad hoc basis. Here we provide an unbiased assessment of the effectiveness of eight of these methods at detecting bursts in a range of spike trains. We suggest a list of features that an ideal burst detection technique should possess and use synthetic data to assess each method in regard to these properties. We further employ each of the methods to reanalyze microelectrode array (MEA) recordings from mouse retinal ganglion cells and examine their coherence with bursts detected by a human observer. We show that several common burst detection techniques perform poorly at analyzing spike trains with a variety of properties. We identify four promising burst detection techniques, which are then applied to MEA recordings of networks of human induced pluripotent stem cell-derived neurons and used to describe the ontogeny of bursting activity in these networks over several months of development. We conclude that no current method can provide "perfect" burst detection results across a range of spike trains; however, two burst detection techniques, the MaxInterval and logISI methods, outperform compared with others. We provide recommendations for the robust analysis of bursting activity in experimental recordings using current techniques.

Detecting pairwise correlations in spike trains: an objective comparison of methods and application to the study of retinal waves.

Correlations in neuronal spike times are thought to be key to processing in many neural systems. Many measures have been proposed to summarize these correlations and of these the correlation index is widely used and is the standard in studies of spontaneous retinal activity. We show that this measure has two undesirable properties: it is unbounded above and confounded by firing rate. We list properties needed for a measure to fairly quantify and compare correlations and we propose a novel measure of correlation-the spike time tiling coefficient. This coefficient, the correlation index, and 33 other measures of correlation of spike times are blindly tested for the required properties on synthetic and experimental data. Based on this, we propose a measure (the spike time tiling coefficient) to replace the correlation index. To demonstrate the benefits of this measure, we reanalyze data from seven key studies, which previously used the correlation index to investigate the nature of spontaneous activity. We reanalyze data from ß2(KO) and ß2(TG) mutants, mutants lacking connexin isoforms, and also the age-dependent changes in wild-type and ß2(KO) correlations. Reanalysis of the data using the proposed measure can significantly change the conclusions. It leads to better quantification of correlations and therefore better inference from the data. We hope that the proposed measure will have wide applications, and will help clarify the role of activity in retinotopic map formation.

Estimating the location and size of retinal injections from orthogonal images of an intact retina.

BACKGROUND: To study the mapping from the retina to the brain, typically a small region of the retina is injected with a dye, which then propagates to the retina's target structures. To determine the location of the injection, usually the retina is dissected out of the eye, flattened and then imaged, causing tears and stretching of the retina. The location of the injection is then estimated from the image of the flattened retina. Here we propose a new method that avoids dissection of the retina. RESULTS: We have developed IntactEye, a software package that uses two orthogonal images of the intact retina to locate focal injections of a dye. The two images are taken while the retina is still inside the eye. This bypasses the dissection step, avoiding unnecessary damage to the retina, and speeds up data acquisition. By using the native spherical coordinates of the eye, we avoid distortions caused by interpreting a curved structure in a flat coordinate system. Our method compares well to the projection method and to the Retistruct package, which both use the flattened retina as a starting point. We have tested the method also on synthetic data, where the injection location is known. Our method has been designed for analysing mouse retinas, where there are no visible landmarks for discerning retinal orientation, but can also be applied to retinas from other species. CONCLUSIONS: IntactEye allows the user to precisely specify the location and size of a retinal injection from two orthogonal images taken of the eye. We are solving the abstract problem of locating a point on a spherical object from two orthogonal images, which might have applications outside the field of neuroscience.

Following the ontogeny of retinal waves: pan-retinal recordings of population dynamics in the neonatal mouse.

The immature retina generates spontaneous waves of spiking activity that sweep across the ganglion cell layer during a limited period of development before the onset of visual experience. The spatiotemporal patterns encoded in the waves are believed to be instructive for the wiring of functional connections throughout the visual system. However, the ontogeny of retinal waves is still poorly documented as a result of the relatively low resolution of conventional recording techniques. Here, we characterize the spatiotemporal features of mouse retinal waves from birth until eye opening in unprecedented detail using a large-scale, dense, 4096-channel multielectrode array that allowed us to record from the entire neonatal retina at near cellular resolution. We found that early cholinergic waves propagate with random trajectories over large areas with low ganglion cell recruitment. They become slower, smaller and denser when GABAA signalling matures, as occurs beyond postnatal day (P) 7. Glutamatergic influences dominate from P10, coinciding with profound changes in activity dynamics. At this time, waves cease to be random and begin to show repetitive trajectories confined to a few localized hotspots. These hotspots gradually tile the retina with time, and disappear after eye opening. Our observations demonstrate that retinal waves undergo major spatiotemporal changes during ontogeny. Our results support the hypotheses that cholinergic waves guide the refinement of retinal targets and that glutamatergic waves may also support the wiring of retinal receptive fields.

MEA-NAP compares microscale functional connectivity, topology, and network dynamics in organoid or monolayer neuronal cultures.

Microelectrode array (MEA) recordings are commonly used to compare firing and burst rates in neuronal cultures. MEA recordings can also reveal microscale functional connectivity, topology, and network dynamics-patterns seen in brain networks across spatial scales. Network topology is frequently characterized in neuroimaging with graph theoretical metrics. However, few computational tools exist for analyzing microscale functional brain networks from MEA recordings. Here, we present a MATLAB MEA network analysis pipeline (MEA-NAP) for raw voltage time-series acquired from single- or multi-well MEAs. Applications to 3D human cerebral organoids or 2D human-derived or murine cultures reveal differences in network development, including topology, node cartography, and dimensionality. MEA-NAP incorporates multi-unit template-based spike detection, probabilistic thresholding for determining significant functional connections, and normalization techniques for comparing networks. MEA-NAP can identify network-level effects of pharmacologic perturbation and/or disease-causing mutations and, thus, can provide a translational platform for revealing mechanistic insights and screening new therapeutic approaches.

Parasol cell mosaics are unlikely to drive the formation of structured orientation maps in primary visual cortex.

The receptive fields of on- and off-center parasol cell mosaics independently tile the retina to ensure efficient sampling of visual space. A recent theoretical model represented the on- and off-center mosaics by noisy hexagonal lattices of slightly different density. When the two lattices are overlaid, long-range Moiré interference patterns are generated. These Moiré interference patterns have been suggested to drive the formation of highly structured orientation maps in visual cortex. Here, we show that noisy hexagonal lattices do not capture the spatial statistics of parasol cell mosaics. An alternative model based upon local exclusion zones, termed as the pairwise interaction point process (PIPP) model, generates patterns that are statistically indistinguishable from parasol cell mosaics. A key difference between the PIPP model and the hexagonal lattice model is that the PIPP model does not generate Moiré interference patterns, and hence stimulated orientation maps do not show any hexagonal structure. Finally, we estimate the spatial extent of spatial correlations in parasol cell mosaics to be only 200-350 µm, far less than that required to generate Moiré interference. We conclude that parasol cell mosaics are too disordered to drive the formation of highly structured orientation maps in visual cortex.

Sepsis-3 criteria in AmsterdamUMCdb: open-source code implementation.

Sepsis is a major healthcare problem with substantial mortality and a common reason for admission to the intensive care unit (ICU). For this reason, the management of sepsis is an important area of ICU research. A number of large-scale, freely-accessible ICU databases are available for observational research and the robust identification of septic patients in such data sets is crucial for research purposes, particularly for comparative studies between critical care sub-populations which may vary around the world. However, data structures are poorly standardised due to inevitable variances in clinical electronic health record system vendor and implementation as well as research database design choices. Robust and well-documented cohort selection (such as patients with sepsis) is crucial for reproducible research. In this work, we operationalise the Sepsis-3 definition on the AmsterdamUMCdb, a recently published large European ICU database, publishing open-access code for wider use by critical care researchers.

CODECHECK: an Open Science initiative for the independent execution of computations underlying research articles during peer review to improve reproducibility.

The traditional scientific paper falls short of effectively communicating computational research.  To help improve this situation, we propose a system by which the computational workflows underlying research articles are checked. The CODECHECK system uses open infrastructure and tools and can be integrated into review and publication processes in multiple ways. We describe these integrations along multiple dimensions (importance, who, openness, when). In collaboration with academic publishers and conferences, we demonstrate CODECHECK with 25 reproductions of diverse scientific publications. These CODECHECKs show that asking for reproducible workflows during a collaborative review can effectively improve executability. While CODECHECK has clear limitations, it may represent a building block in Open Science and publishing ecosystems for improving the reproducibility, appreciation, and, potentially, the quality of non-textual research artefacts. The CODECHECK website can be accessed here: https://codecheck.org.uk/.

LL5beta: a regulator of postsynaptic differentiation identified in a screen for synaptically enriched transcripts at the neuromuscular junction.

In both neurons and muscle fibers, specific mRNAs are concentrated beneath and locally translated at synaptic sites. At the skeletal neuromuscular junction, all synaptic RNAs identified to date encode synaptic components. Using microarrays, we compared RNAs in synapse-rich and -free regions of muscles, thereby identifying transcripts that are enriched near synapses and that encode soluble membrane and nuclear proteins. One gene product, LL5beta, binds to both phosphoinositides and a cytoskeletal protein, filamin, one form of which is concentrated at synaptic sites. LL5beta is itself associated with the cytoplasmic face of the postsynaptic membrane; its highest levels border regions of highest acetylcholine receptor (AChR) density, which suggests a role in "corraling" AChRs. Consistent with this idea, perturbing LL5beta expression in myotubes inhibits AChR aggregation. Thus, a strategy designed to identify novel synaptic components led to identification of a protein required for assembly of the postsynaptic apparatus.

Burst-time-dependent plasticity robustly guides ON/OFF segregation in the lateral geniculate nucleus.

Spontaneous retinal activity (known as "waves") remodels synaptic connectivity to the lateral geniculate nucleus (LGN) during development. Analysis of retinal waves recorded with multielectrode arrays in mouse suggested that a cue for the segregation of functionally distinct (ON and OFF) retinal ganglion cells (RGCs) in the LGN may be a desynchronization in their firing, where ON cells precede OFF cells by one second. Using the recorded retinal waves as input, with two different modeling approaches we explore timing-based plasticity rules for the evolution of synaptic weights to identify key features underlying ON/OFF segregation. First, we analytically derive a linear model for the evolution of ON and OFF weights, to understand how synaptic plasticity rules extract input firing properties to guide segregation. Second, we simulate postsynaptic activity with a nonlinear integrate-and-fire model to compare findings with the linear model. We find that spike-time-dependent plasticity, which modifies synaptic weights based on millisecond-long timing and order of pre- and postsynaptic spikes, fails to segregate ON and OFF retinal inputs in the absence of normalization. Implementing homeostatic mechanisms results in segregation, but only with carefully-tuned parameters. Furthermore, extending spike integration timescales to match the second-long input correlation timescales always leads to ON segregation because ON cells fire before OFF cells. We show that burst-time-dependent plasticity can robustly guide ON/OFF segregation in the LGN without normalization, by integrating pre- and postsynaptic bursts irrespective of their firing order and over second-long timescales. We predict that an LGN neuron will become ON- or OFF-responsive based on a local competition of the firing patterns of neighboring RGCs connecting to it. Finally, we demonstrate consistency with ON/OFF segregation in ferret, despite differences in the firing properties of retinal waves. Our model suggests that diverse input statistics of retinal waves can be robustly interpreted by a burst-based rule, which underlies retinogeniculate plasticity across different species.

A multi-component model of the developing retinocollicular pathway incorporating axonal and synaptic growth.

During development, neurons extend axons to different brain areas and produce stereotypical patterns of connections. The mechanisms underlying this process have been intensively studied in the visual system, where retinal neurons form retinotopic maps in the thalamus and superior colliculus. The mechanisms active in map formation include molecular guidance cues, trophic factor release, spontaneous neural activity, spike-timing dependent plasticity (STDP), synapse creation and retraction, and axon growth, branching and retraction. To investigate how these mechanisms interact, a multi-component model of the developing retinocollicular pathway was produced based on phenomenological approximations of each of these mechanisms. Core assumptions of the model were that the probabilities of axonal branching and synaptic growth are highest where the combined influences of chemoaffinity and trophic factor cues are highest, and that activity-dependent release of trophic factors acts to stabilize synapses. Based on these behaviors, model axons produced morphologically realistic growth patterns and projected to retinotopically correct locations in the colliculus. Findings of the model include that STDP, gradient detection by axonal growth cones and lateral connectivity among collicular neurons were not necessary for refinement, and that the instructive cues for axonal growth appear to be mediated first by molecular guidance and then by neural activity. Although complex, the model appears to be insensitive to variations in how the component developmental mechanisms are implemented. Activity, molecular guidance and the growth and retraction of axons and synapses are common features of neural development, and the findings of this study may have relevance beyond organization in the retinocollicular pathway.

From random to regular: Variation in the patterning of retinal mosaics.

The various types of retinal neurons are each positioned at their respective depths within the retina where they are believed to be assembled as orderly mosaics, in which like-type neurons minimize proximity to one another. Two common statistical analyses for assessing the spatial properties of retinal mosaics include the nearest neighbor analysis, from which an index of their "regularity" is commonly calculated, and the density recovery profile derived from autocorrelation analysis, revealing the presence of an exclusion zone indicative of anti-clustering. While each of the spatial statistics derived from these analyses, the regularity index and the effective radius, can be useful in characterizing such properties of orderly retinal mosaics, they are rarely sufficient for conveying the natural variation in the self-spacing behavior of different types of retinal neurons and the extent to which that behavior generates uniform intercellular spacing across the mosaic. We consider the strengths and limitations of these and other spatial statistical analyses for assessing the patterning in retinal mosaics, highlighting a number of misconceptions and their frequent misuse. Rather than being diagnostic criteria for determining simply whether a population is "regular," they should be treated as descriptive statistics that convey variation in the factors that influence neuronal positioning. We subsequently apply multiple spatial statistics to the analysis of eight different mosaics in the mouse retina, demonstrating conspicuous variability in the degree of patterning present, from essentially random to notably regular. This variability in patterning has both a developmental as well as a functional significance, reflecting the rules governing the positioning of different types of neurons as the architecture of the retina is assembled, and the distinct mechanisms by which they regulate dendritic growth to generate their characteristic coverage and connectivity.

Burst Detection Methods.

'Bursting', defined as periods of high-frequency firing of a neuron separated by periods of quiescence, has been observed in various neuronal systems, both in vitro and in vivo. It has been associated with a range of neuronal processes, including efficient information transfer and the formation of functional networks during development, and has been shown to be sensitive to genetic and pharmacological manipulations. Accurate detection of periods of bursting activity is thus an important aspect of characterising both spontaneous and evoked neuronal network activity. A wide variety of computational methods have been developed to detect periods of bursting in spike trains recorded from neuronal networks. In this chapter, we review several of the most popular and successful of these methods.

Functional characterization of human pluripotent stem cell-derived cortical networks differentiated on laminin-521 substrate: comparison to rat cortical cultures.

Human pluripotent stem cell (hPSC)-derived neurons provide exciting opportunities for in vitro modeling of neurological diseases and for advancing drug development and neurotoxicological studies. However, generating electrophysiologically mature neuronal networks from hPSCs has been challenging. Here, we report the differentiation of functionally active hPSC-derived cortical networks on defined laminin-521 substrate. We apply microelectrode array (MEA) measurements to assess network events and compare the activity development of hPSC-derived networks to that of widely used rat embryonic cortical cultures. In both of these networks, activity developed through a similar sequence of stages and time frames; however, the hPSC-derived networks showed unique patterns of bursting activity. The hPSC-derived networks developed synchronous activity, which involved glutamatergic and GABAergic inputs, recapitulating the classical cortical activity also observed in rodent counterparts. Principal component analysis (PCA) based on spike rates, network synchronization and burst features revealed the segregation of hPSC-derived and rat network recordings into different clusters, reflecting the species-specific and maturation state differences between the two networks. Overall, hPSC-derived neural cultures produced with a defined protocol generate cortical type network activity, which validates their applicability as a human-specific model for pharmacological studies and modeling network dysfunctions.