Predicting species emergence in complex networks

by Omer Markovitch

16:00 (40 min) in CT 7.01

Lognormal distributions are abundant in nature, e.g. distributions of earth's crust chemical and radioactive composition, latent periods of infectious diseases, abundance of bacteria on plants, olfactory receptors binding affinities, etc., have been shown to fit a lognormal distribution. Lognormal networks are graphs whose edge-weights follow a lognormal distribution. Here we utilise lognormal networks to analyse the emergence of "species” within an Artificial Life model of prebiotic evolution, based on the graded autocatalysis replication domain (GARD) simulator for prebiotic molecular assemblies. The temporal evolution of these assemblies is stochastically determined by a rates matrix - typically drawn from a lognormal distribution - that governs the likelihood that a given molecule joins or leaves an assembly.

We re-interpreted here the rates matrix as a network and we analyzed its community structure. We asked whether communities are related (and how) to the evolved species under stochastic dynamic GARD and found that the derived communities correspond well to the species that emerge from the prebiotic evolution simulations. Importantly, we show that it is possible to predict proto-species emergence without performing any simulations. The analysis developed in this paper may have impact in other areas of Artificial Life, Complex Chemical Systems and Synthetic Biology.