Large scale single nucleotide polymorphism discovery in unsequenced genomes using second generation high throughput sequencing technology: applied to turkey

A total of 100 million 36 bp reads were generated, representing approximately 5-6% (~62 Mbp) of the turkey genome, with an estimated sequence depth of 58. Reads consisting of bases called with less than 1% error probability were selected and assembled into contigs. Subsequently, high throughput discovery of nucleotide variation was performed using sequences with more than 90% reliability by using the assembled contigs that were 50 bp or longer as the reference sequence. We identified more than 7,500 SNPs with a high probability of representing true nucleotide variation in turkeys. Increasing the reference genome by adding publicly available turkey BAC-end sequences increased the number of SNPs to over 11,000. A comparison with the sequenced chicken genome indicated that the assembled turkey contigs were distributed uniformly across the turkey genome. Genotyping of a representative sample of 340 SNPs resulted in a SNP conversion rate of 95%. The correlation of the minor allele count (MAC) and observed minor allele frequency (MAF) for the validated SNPs was 0.69.We provide an efficient and cost-effective approach for the identification of thousands of high quality SNPs in species currently lacking a sequenced genome and applied this to turkey. The methodology addresses a random fraction of the genome, resulting in an even distribution of SNPs across the targeted genome.The scalability and availability of highly automated genotyping assays for single nucleotide polymorphisms (SNPs) has made the SNP a popular marker in genetic linkage and association studies in a variety of species. In humans, large-scale identification and characterization has resulted in a repository of over 14 million SNPs [1] that are now being used in whole genome association studies to identify genes involved in complex genetic traits [2-6]. The availability of a high quality reference genome sequence and resources to perform low coverage resequencing on a few individuals are prerequisites for the