Figure 2

Prefiltering step in kClust. Prefiltering algorithm: For each k-mer in the query (k = 6), a list of similar k-mers and their BLOSUM62 similarity scores is generated (blue frame). For each such k-mer (red), a pointer to a list of representative sequences containing this k-mer is looked up in an array (index table). The score S of each sequence in that list is increased by the similarity score. After all k-mers in the query have been processed, array S contains for each representative sequence the sum of k-mer similarity scores.