Input is a set of DNA sequences adjacent or around a functional site
plus a sequence motif defined by a consensus sequence or a weight
matrix. The input sequence motif serves as an
initial model for starting the optimization process. The output
is a complete weight matrix-based description of a sequence motif
maximizing a quantitative criterion of local over-representation,
including the optimized border positions and cut-off value.
A detailed description of the method can be found here