Model Checking Tap Withdrawal in C. Elegans

Md. Ariful Islam, Richard DeFrancisco, Chuchu Fan, Radu Grosu, Sayan Mitra, Scott A. Smolka

Abstract

This study presents the first formal verification of a biologically realistic, nonlinear ordinary differential equation (ODE) model of a neural circuit in a multicellular organism (C. elegans). By applying reachability analysis and "discrepancy functions," the researchers successfully identified the specific ranges of synaptic parameters that trigger the worm's "tap withdrawal" reflex, bridging the gap between formal methods and systems biology.

Introduction

In systems biology, neural circuits are typically modeled using complex systems of nonlinear ODEs. While these models are powerful, they are traditionally analyzed through point-simulations with fixed parameters. However, biological systems are inherently variable—parameters like synaptic conductance or membrane capacitance differ across individuals and environments. Traditional simulation cannot prove a property holds for an entire range of possible values; it can only describe behavior at a single point.

This research addresses the need for formal verification in neurology: a method to mathematically guarantee which parameter ranges lead to specific behaviors—such as a worm reversing its movement—versus others, like accelerating or failing to respond.

The Tap Withdrawal Circuit Model

The authors focus on the Tap Withdrawal (TW) circuit of the roundworm Caenorhabditis elegans. They utilize a 9-dimensional nonlinear ODE model based on the work of Wicks et al. (1996), which describes the membrane potentials of sensory and inter-neurons involved in the reflex.

The neural circuit for tap withdrawal in C. elegans, showing the connections between sensory neurons and interneurons.

Methodology

To analyze this complex nonlinear system, the researchers employed several advanced formal methods:

  1. Reachability Analysis: Instead of single-point simulations, they compute "reach tubes"—volumes in state space that contain all possible trajectories for a given set of initial conditions and parameter ranges.
  2. Local Discrepancy Functions: To make reachability analysis tractable, they used a technique to automatically compute the rate at which nearby trajectories converge or diverge (discrepancy), allowing for tighter over-approximations of the system's behavior.
  3. $δ$-Refinement: If the analysis cannot definitively prove a behavior (e.g., the reach tube overlaps both "reversal" and "no response" zones), the algorithm iteratively partitions the parameter space into smaller pieces until a conclusion is reached.

Results and Validation

The analysis successfully identified parameter ranges for all three observed behaviors: reversal, acceleration, and no response. The researchers focused on varying the conductance of gap junctions (electrical synapses), finding them to be the primary "control knobs" for the circuit's output.

To validate the model, the team compared their results against experimental in vivo data from laser ablation studies, where specific neurons are destroyed to observe behavioral changes.

Experimental behavioral responses across various neuron ablation groups, used as a benchmark for model validation.

In the control group (no neurons removed), the model proved that reversal is the dominant behavior, matching the 98% occurrence rate seen in live worms. For most ablation groups, such as PLM- or ALM,DVA-, the model's predicted dominant behaviors matched experimental trends. One notable exception was the ALM- group; the model predicted "reversal" while experiments showed "acceleration." This discrepancy, also noted by the original model's creators, suggests that formal analysis is effective at identifying limitations in the underlying biological model itself.

Discussion and Future Work

This work demonstrates that tools originally designed for verifying hardware and software can be adapted to handle the nonlinearities of biological neural circuits. Unlike previous studies providing single estimated values, this work provides guaranteed parameter ranges.

However, formal verification remains computationally expensive; verifying a single property in a 1-D parameter space currently takes approximately one hour. Future work will focus on scaling these methods to higher dimensions to include more parameters, such as membrane capacitance and leakage current, and parallelizing algorithms to handle the complexity of larger biological systems.

Summary updated

Abstract

We present what we believe to be the first formal verification of a biologically realistic (nonlinear ODE) model of a neural circuit in a multicellular organism: Tap Withdrawal (TW) in \emph{C. Elegans}, the common roundworm. TW is a reflexive behavior exhibited by \emph{C. Elegans} in response to vibrating the surface on which it is moving; the neural circuit underlying this response is the subject of this investigation. Specifically, we perform reachability analysis on the TW circuit model of Wicks et al. (1996), which enables us to estimate key circuit parameters. Underlying our approach is the use of Fan and Mitra's recently developed technique for automatically computing local discrepancy (convergence and divergence rates) of general nonlinear systems. We show that the results we obtain are in agreement with the experimental results of Wicks et al. (1995). As opposed to the fixed parameters found in most biological models, which can only produce the predominant behavior, our techniques characterize ranges of parameters that produce (and do not produce) all three observed behaviors: reversal of movement, acceleration, and lack of response.
BibTeX
@inproceedings{islam2015model,
  author = {Md. Ariful Islam and Richard DeFrancisco and Chuchu Fan and Radu Grosu and Sayan Mitra and Scott A. Smolka},
  title = {{Model Checking Tap Withdrawal in C. Elegans}},
  booktitle = {HSB},
  year = {2015},
  doi = {10.1007/978-3-319-26916-0\_11},
}
Figure