Inference of molecular homology and sequence alignment by direct optimization

Abstract

That homologies exist and can be recognized and discovered is of central importance to comparative and evolutionary biology. Traditionally, homology assessment for both molecular and morphological data has been treated as a two-step process involving the generation of a proposition of homology and then the evaluation of that hypothesis through the test of congruence in a phylogenetic analysis. An alternative phylogenetic method, direct optimization, combines these into a one-step process in which positional homology and cladograms are co-estimated simultaneously. Here we examine the use of the term “homology” as it is applied to molecular data, and critique the concept of molecular homology one must accept if cladograms are to be interpreted as phylogenetic estimates. The test of congruence alone cannot be used to determine homologous (historically identical) features, and character analysis and the proposition of primary homology hypotheses is the only procedure that can identify correspondences between features justified as retaining phylogenetic information. The practical ramifications for phylogenetic analysis, such as data exclusion and the interpretation of tree-like diagrams as phylogenies, are discussed.