Computing Genome Edit Distances using Domain-Independent Planning

Abstract

The use of planning for computing genome edit distances was suggested by Erdem and Tillier in 2005, but to date there has been no study of how well domain-independent planners are able to solve this problem. This paper reports on experiments with several PDDL formulations of the problem, using several state-of-the-art planners. The main observations are, first, that the problem formulation that is easiest for planners to deal with is not the obvious one, and, second, that plan quality � in particular consistent and assured plan quality � remains the biggest challenge.