Combining Gene-Finding Programs by Using Dempster-Shafer Theory of Evidence for Gene Prediction

Although the computational gene-finding programs have been greatly improved in recent years, our ultimate goal is far to be met. Even the best program cannot be used automatically to identify genes and other genomic elements. Fortunately, there are only a tiny number of exons completely missed by all programs. Therefore the combination of predictions from the gene-finding programs is a convenient way to improve gene prediction. In this paper we motivate the use of the Dempster-Shafer theory of evidence as an appropriate theory for modelling combination of gene predictions, and give the mathematical framework for combining gene predictions of gene-finding programs by using Dempster-Shafer combination rule. The most notable strength of our method lies in the fact that we can combine results of multiple gene-finding programs, in which each of them has provided reliable exon scores. In comparison with an individual program, our method proves an improvement of predictive accuracy both on nucleotide level and exon level, the exon level in particular