online read us now
Paper details
Number 4 - December 2009
Volume 19 - 2009
Some remarks on evaluating the quality of the multiple sequence alignment based on the BAliBASE benchmark
Jacek Błażewicz, Piotr Formanowicz, Paweł Wojciechowski
Abstract
BAliBASE is one of the most widely used benchmarks for multiple sequence alignment programs. The accuracy of alignment methods is measured by bali score—an application provided together with the database. The standard accuracy measures are the Sum of Pairs (SP) and the Total Column (TC). We have found that, for non-core block columns, results calculated by bali score are different from those obtained on the basis of the formal definitions of the measures. We do not claim that one of these measures is better than the other, but they are definitely different. Such a situation can be the source of confusion when alignments obtained using various methods are compared. Therefore, we propose a new nomenclature for the measures of the quality of multiple sequence alignments to distinguish which one was actually calculated. Moreover, we have found that the occurrence of a gap in some column in the first sequence of the reference alignment causes column discarding.
Keywords
multiple sequence alignment, reference alignment, alignment accuracy