Abstract
One fascinating problem in Bioinformatics research area is to denote gene structure from genomic sequences. Some methods had been published and proved useful, but they all consider the information from one sequence only. The newest development shall be cross-species sequence comparison, which take two or more sequences into consideration. We take the assumption that important functional elements tend to be strongly conserved than other intergenic sequences under the evolution pressure. Hence, we introduce a method that combines useful existing software and automates by Perl scripts for detecting protein-coding regions. This strategy has three key parts: sequence alignment, the KA/KS ratio test, and boundary determinant. It is simple and powerful to implement, and easy to extend in the future. A test dataset of selected orthologous genes is included in the performance test. The results show we have good performance, and do find most exon boundaries correct. The method shall be furthermore established as an automated data analysis system. An initial web page was constructed at http://nekrut.uchicago.edu/eev/ as the evolutionary exon validation tool.