Efficient finding of Nearly-ALL local alignments (the NALL method) [2] that score above a cutoff or probability threshold, between a sequence and a database. This means if two proteins have more than one common region, most regions are reported. Effectively, this is like BLAST [1] but with gapped alignments.
Efficient implementation of the Smith-Waterman Algorithm - this returns the highest scoring local alignment between two sequences including gaps where necessary. The program is approximately a factor of three faster than sssearch.
Estimation of the significance of the local alignments. An empirical method is used which takes into account the alignment score and the alignment length. This has the effect of pushing unusually high scoring, but short alignments higher up the hit list.
Comparison of all pairs of sequences in a set using either the Smith-Waterman, or NALL methods.