The field of promoter prediction holds almost limitless potential, but has had very limited success. Instead of analyzing DMA sequence homology, we propose a novel method of eucaryotic promoter prediction based on the physical properties of DNA which arise from the local sequence. Using computer simulations with a nonlinear mathematical model, we predict the localized opening profile of a region of DNA. For several sample eucaryotic promoters, we have demonstrated that the dominant preferential opening positions predicted by the model (and verified by S1 nuclease digestion assays) correlate well with the experimentally-determined transcriptional start sites or major regulatory sites of the gene promoters. We hypothesize that these opening profiles can be applied more generally in seeking out novel gene promoters and transcriptionally significant sites in genomic DNA. Here we propose to further validate the use of nonlinear mathematical models to predict opening profiles as indicators for eukaryotic promoter prediction using known gene core promoters with experimentally-determined transcriptional start sites. We will also seek to apply the computational promoter prediction method in proof-of-concept studies on genes with unidentified promoters and transcriptional start sites. Finally, we plan to develop numerical techniques that allow application of our promoter prediction model on a genomic scale. The simulation-based analysis of DNA opening profiles shows great potential in the prediction of human gene promoters. This method is superior to previous prediction models in that it examines sequence-derived physical properties of DNA rather than sequence homology, however, further investigation is necessary to evaluate and expand the limits of applicability of this method. One of the strongest advantages of the computational model is that it can be used to evaluate opening profiles for any sequence of eukaryotic DNA with very little cost.
Showing the most recent 10 out of 16 publications
