A Multiobjective Evolutionary Fuzzy System for Promoter Discovery in E. coli

In this contribution, the biological problem of extracting promoters (composed of two nucleotide sequences, TTGACA and TATAAT, separated by among 15 and 22 pairs of bases) from E. coli DNA sequences is tackled. Classical approaches for this problem, based on considering probabilistic models of the promoter motifs, fail at performing accurate predictions due to the difficulty of properly integrating the modeled sub-motifs because of the uncertainty existing in the distance between them. However, our methodology solves this problem by applying a multiobjective evolutionary algorithm to extract the promoters, thus being able to discover promoters where the sub-motifs are located at different distances. As we consider the sub-motifs to be modeled by fuzzy logic tools, and evolutionary algorithms are also used to tune these fuzzy models, the resulting technique becomes a multiobjective evolutionary fuzzy system. Some experiments to extract previously known and unknown promoters from E. coli DNA sequences are reported to show its good performance when compared to classical techniques. This method is available for public use in http://gps-tools.wustl.edu.