I too would suggest taking a look at the PDL...

it's whole purpose is to provide an efficient, perl based mechanism for efficiently manipulating and operating over large data sets...

you may particularly be interested in some of the bioinformatics resources available on the net where you will likely find existing code in your area of research...www.bioperl.com

the native PDL data structure is commonly called a "piddle" and can represent something as simple as a two-dimensional array or n-dimensional data cube...you may have to rethink many of your algorithms though, primarily because PDL's builtin functions/operators take care of iterating over the dataset for you... hope this helps... .