Wise2 is a package focused on comparisons of biopolymers, commonly DNA sequence and protein sequence. There are many other packages which do this, probably the best known being BLAST package (from NCBI) and the Fasta package (from Bill Pearson). There are other packages, such as the HMMER package (Sean Eddy) or SAM package (UC Santa Cruz) focused on hidden Markov models (HMMs) of biopolymers.
Wise2's is now a collection of algorithms which generally differ from the usualy, ``standard'' bioinformatics comparison methods. Probably the most used algorithm in Wise2 is genewise, which is the comparison of DNA sequence at the level of its protein translation. This comparison allows the simultaneous prediction of say gene structure with homology based alignment. However Wise2 has a number of other methods which I hope will also appeal to users - promoterwise for comparing upstream regions, genomewise as a ``protein gene finisher'' tool for combining disparate evidence strands and scanwisep as a fast but sensitive search method.
Wise2, although implemented in C makes heavy use of the Dynamite code generating language. Dynamite was written for this project, by myself (Ewan Birney). There is a separate documentation for Dynamite found in the same place as this file; to be honest noone except me can be expected to use Dynamite; it is a cranky, stupid, badly written code generating language with about two thirds of its code base dedicated to dynamic programming, and the remaining third a rather bad primitive object based system. It does however fit me like a glove, and makes my own personal code efficiency impressive.
Wise2 has had a varied life. It started off as a rewrite of my old pairwise and searchwise package (called WiseTools), hence the name Wise2. For a while it looked as if it was going to be pensioned off and live its life out as a rather solid code base for genewise and estwise, but mid 2002 I came back to it for long and complex reasons not worth putting into documentation. This lead to the new methods such as promoterwise and scanwisep which I have high hopes for, though the real test is with real data used by real users...