Progressive multiple alignment based on pyramidal classification and applied to multi-domain proteins
Abstract
In spite of the increasing efficiency of multiple alignment programs, ClustalW still remains the most used algorithm for its good compromise between computing needs and accuracy. We propose to develop a progressive algorithm of the Clustal type with the advantages of a mixed strategy, i.e. combining local and global alignments. For this purpose, we replace the Neighbor-Joining algorithm by a pyramidal classification in order to compute the guide structure. A simple example is given to illustrate this new method.