ASmodeler: Gene modeling of alternative splicing events from genomic alignment of mRNA and ESTs

Namshin Kim1,2, Seokmin Shin2, Sanghyuk Lee
1deepreds@hanmail.net, Division of Molecular Life Sciences, Ewha Womans University, Seoul 120-750, KOREA; 2sshin@snu.ac.kr, School of Chemistry, Seoul National University

ASmodeler is a novel web-based utility to find gene models of alternative splicing (AS) events from genomic alignment of mRNA and ESTs. Alternative splicing is an important mechanism of modulating gene function, and it is directly related to many diseases. We provide a fast and versatile web program for AS modeling.

Input sequences are aligned against the genome map using BLAT program developed by J. Kent at UCSC. Suspicious alignments with noncanonical introns and with tiny exons are corrected using dynamic programming (Sim4). BLAT-Sim4 combination gives fast and reliable genomic alignments. Aligned exons are analyzed in a graph theoretic fashion. Paths along the compatible exons are constructed from the directed acyclic graph (DAG) of exons. Standard depth-first search (DFS) algorithm produces all possible gene models of AS events. mRNA and ESTs are grouped for each gene model as a supporting evidence. ASmodeler is essentially a transcript assembly program as well as an EST clustering utility.

Web-implementation of ASmodeler supports various user options. Many alignment criteria, such as alignment program, method of dealing with multiple hits, percent identity and alignment coverage filtering, are under user control. One can select the best alignment in whole genome or in one chromosome. Input sequences can be aligned against a specific genomic region in various ways. Furthermore, well-known gene predictions (ex. Ensembl, Fgenesh++, Acembly, Genscan and etc) can be included before generating gene models. Three genome maps – human, mouse, and rat – are currently being supported at http://genome.ewha.ac.kr/ASmodeler/.

ASmodeler can be applied as a method of comparative gene modeling. For example, homologous sequences from other species can be included to search for putative AS variants in the human genome.

50-word Abstract: