A set of fixed-length DNA sequences aligned with
respect to a functional site (e.g. a transcription start
site) is scanned in a sliding window in order to determine
the occurrence frequencies of a collection of sequences
motifs contained in a signal sequence collections.
The signal sequence collection may for instance consist
of all tetranucleotides consensus sequences with one
mismatch allowed. The output is a list of signals which
are over- or under represented in particular windows
as compared to a random expectation.
A detailed description of the method can be found here