Enhanced HMM for the Recognition of Sigma70 Promoters in Escherichia coli

In this paper, we propose an enhanced HMM for the recognition of sigma70 promoters in E. coli. HMMs for -10 and -35 boxes have been proposed to model the positional dependency of motifs which is lost in methods based on weight matrices. We also propose to use a set of spacer states sharing the observation densities to achieve the desired spacer duration probability functions. We have conducted two sets of experiments on recognizing promoters and locating DNA binding sites and the proposed method has achieved very promising results in comparison with earlier neural network approaches.