Dynamic response phenotypes and model discrimination in systems and synthetic biology

TL;DR

This paper emphasizes the importance of transient response features in biological systems for understanding network architecture, proposing control-theoretic tools to discriminate models based on dynamic phenotypes rather than steady states.

Contribution

It introduces an input-output framework focusing on transient behaviors, highlighting how motifs like incoherent feedforward generate adaptive responses and how monotonicity constrains model classes.

Findings

Incoherent feedforward motifs produce non-monotonic, adaptive responses.

Monotonic systems impose constraints that can falsify models from transient data.

Periodic forcing and cumulative responses reveal hidden network structures.

Abstract

Biological systems encode function not primarily in steady states, but in the structure of transient responses elicited by time-varying stimuli. Overshoots, biphasic dynamics, adaptation kinetics, fold-change detection, entrainment, and cumulative exposure effects often determine phenotypic outcomes, yet are poorly captured by classical steady-state or dose-response analyses. This paper develops an input-output perspective on such "dynamic phenotypes," emphasizing how qualitative features of transient behavior constrain underlying network architectures independently of detailed parameter values. A central theme is the role of sign structure and interconnection logic, particularly the contrast between monotone systems and architectures containing antagonistic pathways. We show how incoherent feedforward (IFF) motifs provide a simple and recurrent mechanism for generating non-monotonic and adaptive responses across multiple levels of biological organization, from molecular signaling to immune regulation and population dynamics. Conversely, monotonicity imposes sharp impossibility results that can be used to falsify entire classes of models from transient data alone. Beyond step inputs, we highlight how periodic forcing, ramps, and integral-type readouts such as cumulative dose responses offer powerful experimental probes that reveal otherwise hidden structure, separate competing motifs, and expose invariances such as fold-change detection. Throughout, we illustrate how control-theoretic concepts, including monotonicity, equivariance, and input-output analysis, can be used not as engineering metaphors, but as precise mathematical tools for biological model discrimination. Thus we argue for a shift in emphasis from asymptotic behavior to transient and input-driven dynamics as a primary lens for understanding, testing, and reverse-engineering biological networks.