Finding the balance between the mathematical and biological optima in multiple sequence alignment

Abstract

Recent advances in evolutionary modelling and alignment methodology enable alignment of sequences with special features and incorporate structural and functional information. However, our reviewing experience and a recent study by Morrison 1 suggest that these newer methods are under-utilized (especially in the communities of molecular systematics and experimental biology), and the resulting alignments are often curated manually. Most often, no clear biological reasoning is invoked during manual alignment; instead only aesthetic qualities are considered, as measured by eye. Such subjectivity is not consistent with core scientific principles. Although we recognize that methodological problems still exist, computerized alignment methods are currently more realistic and can model a variety of evolutionary mechanisms. We also suggest future directions for the further improvement of automatic alignment methods based upon disconnects of existing methods with underlying biological mechanisms. Alignment as a statement of