Reconstruction of Kauffman networks applying trees

According to Kauffman’s theory [S. Kauffman, The Origins of Order, Self-Organization and Selection in Evolution, Oxford University Press, New York, 1993], enzymes in living organisms form a dynamic network, which governs their activity. For each enzyme the network contains: • a collection of enzymes affecting the enzyme and • a Boolean function prescribing next activity of the enzyme as a function of the present activity of the affecting enzymes.Kauffman’s original pure random structure of the connections was criticized by Barabasi and Albert [A.-L. Barabasi, R. Albert, Emergence of scaling in random networks, Science 286 (1999) 509–512]. Their model was unified with Kauffman’s network by Aldana and Cluzel [M. Aldana, P. Cluzel, A natural class of robust networks, Proc. Natl. Acad. Sci. USA 100 (2003) 8710–8714]. Kauffman postulated that the dynamic character of the network determines the fitness of the organism. If the network is either convergent or chaotic, the chance of survival is lessened. If, however, the network is stable and critical, the organism will proliferate. Kauffman originally proposed a special type of Boolean functions to promote stability, which he called the property canalyzing. This property was extended by Shmulevich et al. [I. Shmulevich, H. Lähdesmäki, E.R. Dougherty, J. Astola, W. Zhang, The role of certain Post classes in Boolean network models of genetic networks, Proc. Natl. Acad. Sci. USA 100 (2003) 10734–10739] using Post classes. Following their ideas, we propose decision tree functions for enzymatic interactions. The model is fitted to microarray data of Cogburn et al. [L.A. Cogburn, W. Wang, W. Carre, L. Rejtő, T.E. Porter, S.E. Aggrey, J. Simon, System-wide chicken DNA microarrays, gene expression profiling, and discovery of functional genes, Poult. Sci. Assoc. 82 (2003) 939–951; L.A. Cogburn, X. Wang, W. Carre, L. Rejtő, S.E. Aggrey, M.J. Duclos, J. Simon, T.E. Porter, Functional genomics in chickens: development of integrated-systems microarrays for transcriptional profiling and discovery of regulatory pathways, Comp. Funct. Genom. 5 (2004) 253–261]. In microarray measurements the activity of clones is measured. The problem here is the reconstruction of the structure of enzymatic interactions of the living organism using microarray data. The task resembles summing up the whole story of a film from unordered and perhaps incomplete collections of its pieces. Two basic ingredients will be used in tackling the problem. In our earlier works [L. Rejtő, G. Tusnády, Evolution of random Boolean NK-models in Tierra environment, in: I. Berkes, E. Csaki, M. Csörgő (Eds.), Limit Theorems in Probability an Statistics, Budapest, vol. II, 2002, pp. 499–526] we used an evolutionary strategy called Tierra, which was proposed by Ray [T.S. Ray, Evolution, complexity, entropy and artificial reality, Physica D 75 (1994) 239–263] for investigating complex systems. Here we apply this method together with the tree–structure of clones found in our earlier statistical analysis of microarray measurements [L. Rejtő, G. Tusnády, Clustering methods in microarrays, Period. Math. Hungar. 50 (2005) 199–221].