Yes, I believe that is the correct data structure. I have used this data structure to count occurrences of dna motifs at various positions. Say, I wanted to count and store the total occurences of A, C, G, T in 1000 nucleotide sequences that are 10 nucleotides long at each position. If you wanted a hash of arrays with the keys being A,C,G,T and the elements being positions 1 through 10, then

$position[10]{A}
[download]

would equal the frequency of A at position 10 if you define position 1 as element 1. ... anyway, yes, this is the data structure that I want to store the clusters in, but I still need to figure out how to write a code to determine those clusters. I am still working on it. I think that I might have figured it out, though it is quite crude.