Entropy maximization and free energy minimization are general physics principles for modeling dynamic systems. Notable examples include modeling decision-making within the brain using the free-energy principle, optimizing the accuracy-complexity trade-off when accessing hidden variables with the information bottleneck principle (Tishby et al. 2000), and navigation in random environments using information maximization (Vergassola et al. 2007). Building on this principle, we propose a new class of bandit algorithms that maximize an approximation to the information of a key variable within the system. To this end, we develop an approximated, analytical physics-based representation of the entropy to forecast the information gain of each action and greedily choose the one with the largest information gain. This method yields strong performances in classical bandit settings. Motivated by its empirical success, we prove its asymptotic optimality for the multi-armed bandit problem with Gaussian rewards. Since it encompasses the system’s properties in a single, global functional, this approach can be efficiently adapted to more complex bandit settings. This calls for further investigation of information maximization approaches for multi-armed bandit problems.

The central nervous system can generate various behaviors, including motor responses, which we can observe through video recordings. Recent advances in gene manipulation, automated behavioral acquisition at scale, and machine learning enable us to causally link behaviors to their underlying neural mechanisms. Moreover, in some animals, such as the Drosophila melanogaster larva, this mapping is possible at the unprecedented scale of single neurons, allowing us to identify the neural microcircuits generating particular behaviors. These high-throughput screening efforts, linking the activation or suppression of specific neurons to behavioral patterns in millions of animals, provide a rich dataset to explore the diversity of nervous system responses to the same stimuli. However, important challenges remain in identifying subtle behaviors, including immediate and delayed responses to neural activation or suppression, and understanding these behaviors on a large scale. We here introduce several statistically robust methods for analyzing behavioral data in response to these challenges: 1) A generative physical model that regularizes the inference of larval shapes across the entire dataset. 2) An unsupervised kernel-based method for statistical testing in learned behavioral spaces aimed at detecting subtle deviations in behavior. 3) A generative model for larval behavioral sequences, providing a benchmark for identifying higher-order behavioral changes. 4) A comprehensive analysis technique using suffix trees to categorize genetic lines into clusters based on common action sequences. We showcase these methodologies through a behavioral screen focused on responses to an air puff, analyzing data from 280 716 larvae across 569 genetic lines.

MOTIVATION: As more behavioural assays are carried out in large-scale experiments on Drosophila larvae, the definitions of the archetypal actions of a larva are regularly refined. In addition, video recording and tracking technologies constantly evolve. Consequently, automatic tagging tools for Drosophila larval behaviour must be retrained to learn new representations from new data. However, existing tools cannot transfer knowledge from large amounts of previously accumulated data. We introduce LarvaTagger, a piece of software that combines a pre-trained deep neural network, providing a continuous latent representation of larva actions for stereotypical behaviour identification, with a graphical user interface to manually tag the behaviour and train new automatic taggers with the updated ground truth. RESULTS: We reproduced results from an automatic tagger with high accuracy, and we demonstrated that pre-training on large databases accelerates the training of a new tagger, achieving similar prediction accuracy using less data. AVAILABILITY AND IMPLEMENTATION: All the code is free and open source. Docker images are also available. See gitlab.pasteur.fr/nyx/LarvaTagger.jl.

Physical and functional constraints on biological networks lead to complex topological patterns across multiple scales in their organization. A particular type of higher-order network feature that has received considerable interest is network motifs, defined as statistically regular subgraphs. These may implement fundamental logical and computational circuits and are referred to as "building blocks of complex networks". Their well-defined structures and small sizes also enable the testing of their functions in synthetic and natural biological experiments. Here, we develop a framework for motif mining based on lossless network compression using subgraph contractions. This provides an alternative definition of motif significance which allows us to compare different motifs and select the collectively most significant set of motifs as well as other prominent network features in terms of their combined compression of the network. Our approach inherently accounts for multiple testing and correlations between subgraphs and does not rely on a priori specification of an appropriate null model. It thus overcomes common problems in hypothesis testing-based motif analysis and guarantees robust statistical inference. We validate our methodology on numerical data and then apply it on synaptic-resolution biological neural networks, as a medium for comparative connectomics, by evaluating their respective compressibility and characterize their inferred circuit motifs.

Numerous models have been developed to account for the complex properties of the random walks of biomolecules. However, when analysing experimental data, conditions are rarely met to ensure model identification. The dynamics may simultaneously be influenced by spatial and temporal heterogeneities of the environment, out-of-equilibrium fluxes and conformal changes of the tracked molecules. Recorded trajectories are often too short to reliably discern such multi-scale dynamics, which precludes unambiguous assessment of the type of random walk and its parameters. Furthermore, the motion of biomolecules may not be well described by a single, canonical random walk model. Here, we develop a two-step statistical testing scheme for comparing biomolecule dynamics observed in different experimental conditions without having to identify or make strong prior assumptions about the model generating the recorded random walks. We first train a graph neural network to perform simulation-based inference and thus learn a rich summary statistics vector describing individual trajectories. We then compare trajectories obtained in different biological conditions using a non-parametric maximum mean discrepancy (MMD) statistical test on their so-obtained summary statistics. This procedure allows us to characterise sets of random walks regardless of their generating models, without resorting to model-specific physical quantities or estimators. We first validate the relevance of our approach on numerically simulated trajectories. This demonstrates both the statistical power of the MMD test and the descriptive power of the learnt summary statistics compared to estimates of physical quantities. We then illustrate the ability of our framework to detect changes in α-synuclein dynamics at synapses in cultured cortical neurons, in response to membrane depolarisation, and show that detected differences are largely driven by increased protein mobility in the depolarised state, in agreement with previous findings. The method provides a means of interpreting the differences it detects in terms of single trajectory characteristics. Finally, we emphasise the interest of performing various comparisons to probe the heterogeneity of experimentally acquired datasets at different levels of granularity (e.g., biological replicates, fields of view, and organelles).

MOTIVATION: Single-molecule localization microscopy allows studying the dynamics of biomolecules in cells and resolving the biophysical properties of the molecules and their environment underlying cellular function. With the continuously growing amount of data produced by individual experiments, the computational cost of quantifying these properties is increasingly becoming the bottleneck of single-molecule analysis. Mining these data requires an integrated and efficient analysis toolbox. RESULTS: We introduce TRamWAy, a modular Python library that features: (i) a conservative tracking procedure for localization data, (ii) a range of sampling techniques for meshing the spatio-temporal support of the data, (iii) computationally efficient solvers for inverse models, with the option of plugging in user-defined functions and (iv) a collection of analysis tools and a simple web-based interface. AVAILABILITY AND IMPLEMENTATION: TRamWAy is a Python library and can be installed with pip and conda. The source code is available at https://github.com/DecBayComp/TRamWAy.

[This corrects the article DOI: 10.1371/journal.pcbi.1004579.].

Stochastic simulations are one of the cornerstones of the analysis of dynamical processes on complex networks, and are often the only accessible way to explore their behavior. The development of fast algorithms is paramount to allow large-scale simulations. The Gillespie algorithm can be used for fast simulation of stochastic processes, and variants of it have been applied to simulate dynamical processes on static networks. However, its adaptation to temporal networks remains non-trivial. We here present a temporal Gillespie algorithm that solves this problem. Our method is applicable to general Poisson (constant-rate) processes on temporal networks, stochastically exact, and up to multiple orders of magnitude faster than traditional simulation schemes based on rejection sampling. We also show how it can be extended to simulate non-Markovian processes. The algorithm is easily applicable in practice, and as an illustration we detail how to simulate both Poissonian and non-Markovian models of epidemic spreading. Namely, we provide pseudocode and its implementation in C++ for simulating the paradigmatic Susceptible-Infected-Susceptible and Susceptible-Infected-Recovered models and a Susceptible-Infected-Recovered model with non-constant recovery rates. For empirical networks, the temporal Gillespie algorithm is here typically from 10 to 100 times faster than rejection sampling.