Towards modular verification of pathways: fairness and assumptions

TL;DR

This paper develops a modular verification framework for biochemical pathways, incorporating fairness assumptions to enable efficient property verification of complex systems by focusing on relevant reactions.

Contribution

It introduces a novel modular verification approach for biochemical pathways that includes fairness considerations, with proven correctness and demonstrated on a MAP kinase cascade model.

Findings

Verification efficiency improved by modular approach

Fairness inclusion is essential for accurate pathway modeling

Framework successfully applied to MAP kinase cascade

Abstract

Modular verification is a technique used to face the state explosion problem often encountered in the verification of properties of complex systems such as concurrent interactive systems. The modular approach is based on the observation that properties of interest often concern a rather small portion of the system. As a consequence, reduced models can be constructed which approximate the overall system behaviour thus allowing more efficient verification. Biochemical pathways can be seen as complex concurrent interactive systems. Consequently, verification of their properties is often computationally very expensive and could take advantage of the modular approach. In this paper we report preliminary results on the development of a modular verification framework for biochemical pathways. We view biochemical pathways as concurrent systems of reactions competing for molecular resources. A modular verification technique could be based on reduced models containing only reactions involving molecular resources of interest. For a proper description of the system behaviour we argue that it is essential to consider a suitable notion of fairness, which is a well-established notion in concurrency theory but novel in the field of pathway modelling. We propose a modelling approach that includes fairness and we identify the assumptions under which verification of properties can be done in a modular way. We prove the correctness of the approach and demonstrate it on the model of the EGF receptor-induced MAP kinase cascade by Schoeberl et al.