Continuous surrogates versus threshold Boolean networks for modeling Arabidopsis ISR gene regulation

2026-07-25

AI Summary

This study investigates the trade-off between predictive accuracy and mechanistic interpretability in modeling induced systemic resistance (ISR) gene regulation in Arabidopsis thaliana. By comparing continuous surrogate models—random forest and multilayer perceptron—trained on raw gene expression data against threshold Boolean networks trained on symbolically binarized data, the work evaluates single-step prediction, recursive multi-step rollout, and interpretability. The analysis reveals a discrepancy between local numerical precision and global qualitative dynamical fidelity: random forest achieves the best single-step performance in the continuous domain (MAE = 1.910), whereas the threshold Boolean network excels in both single-step and recursive trajectory prediction in the binary domain (trajectory accuracy = 1.0), with the multilayer perceptron attaining a recursive accuracy of 0.986. The findings advocate for complementary use of both modeling paradigms to balance predictive power and biological interpretability.

Abstract

Gene regulatory network modeling often requires balancing predictive accuracy and mechanistic interpretability. In this work, we compare continuous surrogate models and a discrete mechanistic model on the same \textit{Arabidopsis thaliana} induced systemic resistance (ISR) dataset, using both the raw continuous gene-expression measurements and their sign-binarized representation. The study considers eight defense-related genes measured over nine time points and evaluates two continuous predictors, Random Forest (RF) regression and a Multi-Layer Perceptron (MLP), against a threshold Boolean network (TBN). The models are assessed using rolling-origin one-step prediction, recursive multi-step rollout, and interpretability analysis. RF achieved the best average one-step numerical performance in the continuous domain, with an MAE of 1.910 and an RMSE of 2.836, compared with 2.089 and 3.106 for the MLP. In the binary domain, the TBN obtained the best average one-step qualitative performance, with a binary accuracy of 0.550 and a Hamming distance of 3.600, compared with 0.500 and 4.000 for RF, and 0.495 and 4.040 for the MLP. In recursive rollout, the TBN exactly reproduced the observed binarized trajectory, while the MLP also showed near-perfect fidelity, with a trajectory binary accuracy of 0.986, and RF accumulated substantially larger deviation, with a trajectory binary accuracy of 0.708. These results highlight that local numerical accuracy and global qualitative dynamical fidelity are not necessarily aligned, and suggest that continuous surrogates and threshold Boolean networks should be viewed as complementary tools for modeling biological regulation.

Problem

• gene regulatory network
• induced systemic resistance
• Boolean network
• continuous surrogate model
• Arabidopsis thaliana

Innovation

• threshold Boolean network
• continuous surrogate models
• gene regulatory network
• induced systemic resistance
• model interpretability