Skip to main content
. 2010 Jul 22;6(7):e1000862. doi: 10.1371/journal.pcbi.1000862

Figure 2. A Divide & Conquer method to clarify transcription factor binding sites.

Figure 2

D&C consists of 4 phases. 1) Divide BS sequences into submotifs by using the possibilistic implementation of the fuzzy clustering method, and then organizing them hierarchically. The submotifs are encoded into PWMs (i.e., Consensus, Meme and AlignAce) and the distances from a TFBS to the RNAP into distributions represented as fuzzy sets. 2) Combine PWMs from submotifs into a multi-classifier. 3) Optimize the accuracy and complexity of the multi-classifier by using genetic algorithms. 4) Fuse different cis-features into fuzzy IF THEN rules, where the antecedents are the conjunction of the individual features, and the consequents are the prediction of a TFBS.