Simultaneous Alignment and Annotation
Annotation is clearly a central problem in comparative genomics: You use observed molecular evolution to make statements about something you can't observe, such as a folding RNA molecule or the protein gene structure in a genome segment. Alignment is more a nuisance problem that must be solved before interesting data analysis can be performed. The last decade has seen the rise of statistical alignment based on stochastic models of insertions, deletions and substitutions. Optimally, alignment and annotation should be solved together and this has been done in several publications for special cases. However, this is difficult to do properly for statistical alignment, but decent ad hoc methods have been proposed.