Probabilistic sequence alignments: realistic models with efficient algorithms

Abstract

Alignment algorithms usually rely on simplified models of gaps for computational efficiency. Based on an isomorphism between alignments and physical helix-coil models, we show in statistical mechanics that alignments with realistic laws for gaps can be computed with fast algorithms. Improved performances of probabilistic alignments with realistic models of gaps are illustrated. Probabilistic and optimization formulations are compared, with potential implications in many fields and perspectives for computationally efficient extensions to Markov models with realistic long-range interactions.