Using Secondary Structure Information to Perform Multiple Alignment

In this paper an approach devised to perform multiple alignment is described, able to exploit any available secondary structure information. In particular, given the sequences to be aligned, their secondary structure (either available or predicted) is used to perform an initial alignment -to be refined by means of locally-scoped operators entrusted with "rearranging" the primary level. Aimed at evaluating both the performance of the technique and the impact of "true" secondary structure information on the quality of alignments, a suitable algorithm has been implemented and assessed on relevant test cases. Experimental results point out that the proposed solution is particularly effective when used to align low similarity protein sequences.