There are two protocols in DS that can make structure based sequence alignment. However it has been said in manual/tutorial page that both of these protocols work when the sequences used have some sequence identity. So, how much sequence identity/sequence similarity should the sequences have in order to get reliable structural alignment? Are these protocols better than other available for example, CLUSTALW (EXPRESSO)? I have sequences that have sequence identity of only 20% - 23%. If the protocols available in DS are good enough to make predictions, which one of the two is better? Also, if they are not better than other available programs, could you suggest which program would be the best in my case.