Automated, predictive, and interpretable inference of Caenorhabditis elegans escape dynamics

Abstract

The roundworm Caenorhabditis elegans exhibits robust escape behavior in response to rapidly rising temperature. The behavior lasts for a few seconds, shows history dependence, involves both sensory and motor systems, and is too complicated to model mechanistically using currently available knowledge. Instead we model the process phenomenologically, and we use the Sir Isaac dynamical inference platform to infer the model in a fully automated fashion directly from experimental data. The inferred model requires incorporation of an unobserved dynamical variable and is biologically interpretable. The model makes accurate predictions about the dynamics of the worm behavior, and it can be used to characterize the functional logic of the dynamical system underlying the escape response. This work illustrates the power of modern artificial intelligence to aid in discovery of accurate and interpretable models of complex natural systems.

Targeted thermal stimulation and high-content phenotyping reveal that the C. elegans escape response integrates current behavioral state and past experience

“…For the experiments described in this work, two stimulus regimes were used: (1) the animals represented in Figs 2, 3, 4, 7 and 8 were stimulated with a 100 ms pulse focused by a 75 mm lens (37.5 mA, 75 mA or 375 mA, as indicated). (2) the animals represented in Figs 5 and 6 were stimulated with a 133 ms 95 mA pulse focused by a 100 mm lens, in order to facilitate comparison with other studies [16,58,59].…”

Learning Everywhere: A Taxonomy for the Integration of Machine Learning and Simulations

“…Using machine learning (modest emphasis) to represent cell (agent) behavior based on data for prediction of cancer cell behavior [17] 5. Automatic inference of a model of the escape response behavior in a roundworm directly from time series data [18] building on [19], [20]. The unknown parameters in a set of ODE's are determined by fitting data in a hierarchical fashion…”

Inferring the structures of signaling motifs from paired dynamic traces of single cells

“…MISC exploits the heterogeneity among identically treated cells by finding one or more signaling motifs that best explain the relationship between the paired upstream signals and downstream responses across all cells in a population. Although many excellent network inference tools have been established [ 22 , 23 ], many of them work on a different type of data, employ averages, or do not allow for hidden nodes. We believe this to be the first tool that infers dynamic networks from paired time-series traces from single cells and allows for hidden nodes.…”