Plant & Animal Genomes XVI Conference

January 12-16, 2008
Town & Country Convention Center
San Diego, CA

Assessing The Feasibility Of 454 Pyrosequencing For Sequencing The Atlantic Salmon Genome

Nicole L Quinn¹ , Natasha Levenkova² , Pascal Bouffard² , William Chow¹ , Tom Jarvie² , Krzysztof P. Lubieniecki¹ , Tim Harkins³ , Brian Desany² , Ben F. Koop⁴ , William S. Davidson¹

¹  Department of Molecular Biology and Biochemistry, Simon Fraser University, 8888 University Drive, Burnaby, British Columbia, V5A 1S6, Canada

²  454 Life Sciences, 20 Commercial St., Branford, CT, 06405, USA

³  Roche Applied Science, 9115 Hauge Rd. Indianapolis, IN, 46250, USA

⁴  Centre for Biomedical Research, University of Victoria, Victoria, British Columbia, V8W 3N5, Canada

The salmonid fishes are of great economic and social importance given that they support commercial and sports fisheries and a large aquaculture industry. In addition, the salmonids are sentinel species for aquatic ecosystems. The current genomic resources for Atlantic salmon include: BAC libraries, a BAC-based physical map, a genetic map with >1,000 markers, >200,000 BAC-end sequences covering ~3.5% of the genome and >436,000 ESTs. These provide a solid foundation for sequencing and annotating the salmon genome. Sanger technology has set the gold standard for sequence quality; however, limitations including cost, labor-intensiveness and speed have fueled the demand for new sequencing approaches. Sequencing projects using new technologies rely heavily on a “guide sequence” to facilitate assembly. The genome of Atlantic salmon is both large (~3x109 bp) and complex (>40% repetitive). This, combined with a lack of a closely related fish reference genome sequence makes sequencing the Atlantic salmon genome challenging. We examined the feasibility of using 454 pyrosequencing to obtain a full sequence of a salmonid genome by sequencing eight pooled BACs belonging to a well-defined minimum tiling path covering ~1Mb. The 454-generated data (average read 248.5 bp) provided ~35x coverage and allowed gene identification. This was enhanced by the addition of 126 BAC-end sequences, which increased the N50 contig size from 11,497 to 13,455 bp. We are currently integrating 454-generated long-read paired-end sequences into the data with the expectation that this will further reduce the number of sequence contigs while increasing their average size.