This tool attempts to improve an existing protein multiple sequence alignment by dividing the alignment into conserved and variable regions using Gblocks. Variable regions are then split into clusters of similar sequences using BLASTClust. Each variable cluster is aligned separately using MAFFT, and then the partial alignments are assembled using MAFFT's profile alignment option. The variable and conserved regions are then reassembled to complete the alignment. The tool currently cannot be used with nucleotide sequences. An existing alignment is currently required to reduce load on the server.