Compositional Analysis of Homeostasis of Gene Networks by Clustering Algorithms

In this work we present a compositional approach to qualitatively analyse homeostasis of gene networks. The problem of analysing homeostasis of gene networks is 2EXPTIME-complete in the sizes of the network specifications. Due to this high complexity of the problem, only small networks consisting of a few genes were successfully analysed. Since the analysis of homeostasis of gene networks is based on the technique of realisability checking of Linear Temporal Logic formulae, we can apply a compositional algorithm devised to mitigate the computational difficulty in realisability problems. For this, we develop a clustering algorithm to divide network specifications in suitable sizes to utilise the compositional algorithm. We report the experimental results of analyses of homeostasis of several gene networks with our proposed method. Our experiments show a fair improvement especially in the analyses of larger networks.