We discuss the problem of optimally "finishing" a partially sequenced, reconstructed DNA segment. At first sight, this appears to be computationally hard. We construct a series of increasingly realistic models for the problem and show that all of these can in fact be solved to optimality in polynomial time, with near-optimal solutions available in linear time. Implementation of our algorithms could result in a substantial efficiency gain for automated DNA sequencing.
Éva Czabarka, Goran Konjevod, Madhav V. Mar