Structural alignment from sequence alone by integration of predicted secondary structure

Protein sequence alignment is at the core of a variety of fundamental tasks such as homology modeling, fold recognition, evolutionary studies and more. Much attention was dedicated over recent years for alignment of sequences in the twilight zone, where pairwise sequence alignment methods (e.g. BLAST) perform poorly. State of the art methods employ sequence profiles, and sometimes meta-sequence information such as predicted secondary structure assignment (SSA) at various stages of fold recognition methods. However, these methods typically use ad-hoc scoring for SSA substitution and do not use the structural information in the initial stages of profile construction from pairwise alignments (E.g. by PSI-Blast).