A Statistical Content-based Method for Promoter Prediction Based on Sequence Segments with Wildcards

A certain kind of context features, sequence segments with wildcards (called IUPAC words), was used for eukaryotic polymerase II promoter prediction successfully by the content-based promoter prediction system PromoterInspector. Inspired by this, a new statistical method for promoter prediction is designed based on these context features, which also incorporates some techniques used in the well-known interpolated Markov chains (IMC) model. As tested on the vertebrate promoter dataset collected by Martin Reese, the performance of this new method is much better than that of PromoterInspector, and is comparative to that of the popular content-based IMC predictors. Combination of this statistical IUPAC-based method with IMC, results in some improvement in discriminative power as compared with the later