Blasting Out of the Box:

Identification of homologous sequences is a fundamental problem in bioinformatics. Increasing the statistical power of detecting distantly related homologs will greatly help biomedical researchers in genome annotation, such as identifying functions and structures of an uncharacterized protein based on a known homolog. We propose an approach to increase the sensitivity of Blast in detecting homologs. Unlike existing methods, this approach relies on reconstructing an ancestor sequence, which gives a biological justification for homology detection based on the first principles of sequence evolution. Our preliminary studies using exploratory, empirical, and theoretical analyses showed the validity and promise of this novel approach, called “REWIND”.