Gene and alternative splicing annotation with AIR
Abstract
Designing effective and accurate tools for identifying the functional and structural elements in a genome remains at the frontier of genome annotation owing to incompleteness and inaccuracy of the data, limitations in the computational models, and shifting paradigms in genomics, such as alternative splicing. We present a methodology for the automated annotation of genes and their alternatively spliced mRNA transcripts based on existing cDNA and protein sequence evidence from the same species or projected from a related species using syntenic mapping information. At the core of the method is the splice graph, a compact representation of a gene, its exons, introns, and alternatively spliced isoforms. The putative transcripts are enumerated from the graph and assigned confidence scores based on the strength of sequence evidence, and a subset of the high-scoring candidates are selected and promoted into the annotation. The method is highly selective, eliminating the unlikely candidates while retaining 98% of the high-quality mRNA evidence in well-formed transcripts, and produces annotation that is measurably more accurate than some evidence-based gene sets. The process is fast, accurate, and fully automated, and combines the traditionally distinct gene annotation and alternative splicing detection processes in a comprehensive and systematic way, thus considerably aiding in the ensuing manual curation efforts.
Footnotes
-
[Supplemental material is available online at www.genome.org and https://panther.appliedbiosystems.com/publications.jsp.]
-
Article and publication are at http://www.genome.org/cgi/doi/10.1101/gr.2889405.
-
↵4 Present address: Department of Computer Science, George Washington University, Washington, DC 20052, USA.
-
↵5 Corresponding author. E-mail florea{at}gwu.edu; fax (240) 453-3324.
-
- Accepted October 14, 2004.
- Received June 14, 2004.
- Cold Spring Harbor Laboratory Press