FMM allows you to find motifs that are discriminatively enriched in a positive sequences set relative to a negative sequences set. The main novelty is that it can learn a “Feature Motif Model” (FMM) representation of the motif, capturing dependencies between different positions through di-nucleotide features (such as: “G at position 3 and T at position 9”). Mono-nucleotide features are also in use, thus the FMM formalism contains the PSSM one. FMMs are represented by a clear and intuitive logo, easily pointing out important di-nucleotide features. The height of a feature in the logo is linear to its expected occurrence.
:: MORE INFORMATION
Sharon & Lubliner et al.,
A Feature-Based Approach to Modeling Protein-DNA Interactions,
PLoS Comput Biol, 4(8) Aug. 2008