Structure-conditioned input-to-state stability for layer-by-layer molecular computations in parallel chemical reaction networks

Abstract

Molecular computation in chemical reaction networks (CRNs) now constitutes a foundational framework for designing programmable biological systems. However, prevailing design methodologies primarily treat parallelism of chemical reactions as a liability, consequently motivating researchers to redirect research focus toward leveraging parallelism to implement layer-by-layer computations of composite functions in coupled mass-action systems (MASs). MASs exhibiting this property are termed composable. Present composability verification for MASs mainly depends on input-to-state stability (ISS) conditions, with structural characteristics of networks remaining underexplored. This paper investigates the structural conditions under which two MASs are composable. By leveraging ISS-Lyapunov functions, we identify a class of CRN architectures, whose reduced systems have zero deficiency, that guarantee composability with other networks. We also extend our conclusions to encompass some CRN architectures possessing nonzero deficiency. Some examples are presented to demonstrate the validity of our theoretical results. Finally, we employ our methods to devise an algorithm for constructing MASs capable of executing specified molecular computations.