An Optimized Genotyping-by-Sequencing Protocol and SNP Calling Pipeline in Diploid and Hexaploid Ipomoea sp

Genotyping-by-sequencing (GBS) is an inexpensive and high-throughput approach for generating thousands, or even millions of molecular markers. However, problems with data quality, including high genotyping error rates and issues of substantial missing data, remain. Here, we present an optimized protocol and SNP calling pipeline that eliminate the source of these problems. The modified protocol utilizes most of the sequenced reads that predominantly map to the nuclear reference genome. It also improves uniformity of read depth across the genome and increases genotyping accuracy. Besides improving technical issues, we have developed a downstream diagnostic tool to eliminate genotyping errors and low quality variants that are attributable to various biological features. One of these features include paralogous sequences, which can account for up to 71.3 % of genes and can erroneously map to a single locus. Consistent with diploid and hexaploid SNP calls, our diagnostic tool and segregation analysis show reliably that errors in two-thirds of the variants might be due to paralogous sequences. This is consistent with expansion of homologous gene families and gene duplication events in large genomes. In addition to SNPs, CNVs (copy number variants) and indels, we effectively capture methylation-based DNA markers, which are crucial for gene regulation. Upon SNP validation and selection of loci that are polymorphic in majority of individuals representing the global diversity of sweetpotato, a SNP array with minimal ascertainment bias can be developed from these high quality variants.