Using Colored Petri Nets to Construct Coalescent Hidden Markov Models: Automatic Translation from Demographic Specifications to Efficient Inference Methods

摘要

Biotechnological improvements over the last decade has made it economically and technologically feasible to collect large DNA sequence data from many closely related species. This enables us to study the detailed evolutionary history of recent speciation and demographics. Sophisticated statistical methods are needed, however, to extract the information that DNA sequences hold, and a limiting factor in this is dealing with the large state space that the ancestry of large DNA sequences spans. Recently a new analysis method, CoalHMMs, has been developed, that makes it computationally feasible to scan full genome sequences - the complete genetic information of a species - and extract genetic histories from this. Applying this methodology, however, requires that the full state space of ancestral histories can be constructed. This is not feasible to do manually, but by applying formal methods such as Petri nets it is possible to build sophisticated evolutionary histories and automatically derive the analysis models needed. In this paper we describe how to use colored stochastic Petri nets to build CoalHMMs for complex demographic scenarios.