A computational approach for gene assembly and exon annotation based on BLAST

Abstract

Accurate gene assembly and precise exon annotations are two of the key goals of human genome project. Existing gene reference sequences and exon annotations are far from perfection. This paper introduces a new greedy algorithm which makes use of mRNA reference sequence and BLAST tools from NCBI (National Center for Biotechnology Information) to effectively assemble and annotate gene structures. Four pipelined components are included in this approach. 1. Blast parser: extract mRNA-DNA local alignment pairs. 2. Chain finder: transform local alignments to spliced alignment. 3. Assembler: assemble multiple DNA sequences into a continuous DNA sequence based on their spliced alignments with a given mRNA sequence. 4. Annotator: resolve exon-intron boundary based on splicing signals. Test results on one sample set of human genes show that gene assembly and exon annotation using the proposed approach is significantly better than contig references from NCBI. The software is available upon request