Fig. 1
An outline of the MP3 framework. a RPS preparation based on sequenced genome from NCBI, operon information retrieved from DOOR, and identified orthologous genes for a target gene using GOST. The promoters of orthologous operons are generated and then are refined to build RPS. b CBR detection by voting strategy and peak finding. The predicted motifs by six tools (short sequences) are mapped back on promoter sequences, and generate score curves. The peaks on the curve are identified as CBR by a peak calling method. c CBR clustering based on a new graph model. r 0, r 1… are CBRs on promoters, which are clustered together as a related CBR set R 1. The motif finding will performed on these clusters (R 1, R 2, …, R t ) again to build motif profiles. d Motif profiles identification and motif width optimization through curve fitting