A New Approach for Gene Annotation Using Unambiguous Sequence Joints

We address the problem of accurate and automatic gene nding and annotation. This paper describes a new cDNA/DNA homology gene annotation algorithm that combines the results of both BLASTN search and spliced alignment. Compared to other programs currently in use, annotation quality is increased through the unambiguous joining of genomic DNA sequences and use of spliced alignment. We also address gene annotation with both non-canonic splice sites and short exons. Our approach has been tested on the Genie learning subset and the human RefSeqs. This approach exhibited sensitivity and specicity of 97 % with the Genie learning subset