SNP discovery and genetic mapping using genotyping by sequencing of whole genome genomic DNA from a pea RIL population

Abstract

Background: Progress in genetics and breeding in pea still suffers from the limited availability of molecular resources. SNP markers that can be identified through affordable sequencing processes, without the need for prior genome reduction or a reference genome to assemble sequencing data would allow the discovery and genetic mapping of thousands of molecular markers. Such an approach could significantly speed up genetic studies and marker assisted breeding for non-model species.

Results: A total of 419,024 SNPs were discovered using HiSeq whole genome sequencing of four pea lines, followed by direct identification of SNP markers without assembly using the discoSnp tool. Subsequent filtering led to the identification of 131,850 highly designable SNPs, polymorphic between at least two of the four pea lines. A subset of 64,754 SNPs was called and genotyped by short read sequencing on a subpopulation of 48 RILs from the cross 'Baccara' x 'PI180693'. This data was used to construct a WGGBS-derived pea genetic map comprising 64,263 markers. This map is collinear with previous pea consensus maps and therefore with the Medicago truncatula genome. Sequencing of four additional pea lines showed that 33 % to 64 % of the mapped SNPs, depending on the pairs of lines considered, are polymorphic and can therefore be useful in other crosses. The subsequent genotyping of a subset of 1000 SNPs, chosen for their mapping positions using a KASP™ assay, showed that almost all generated SNPs are highly designable and that most (95 %) deliver highly qualitative genotyping results. Using rather low sequencing coverages in SNP discovery and in SNP inferring did not hinder the identification of hundreds of thousands of high quality SNPs.

Conclusions: The development and optimization of appropriate tools in SNP discovery and genetic mapping have allowed us to make available a massive new genomic resource in pea. It will be useful for both fine mapping within chosen QTL confidence intervals and marker assisted breeding for important traits in pea improvement.

Fig. 1

Frequency histogram of the number of missing data points in WGGBS of 48 RILs with the 88,851 reliable SNPs that are polymorphic between the ‘Baccara’ and ‘PI180693’ parentals (For example, 13,187, 13,186,and 14,452, were genotyped with 0, 1, or 2 missing data points, respectively among the 48 sequenced RILs)

Fig. 2

Dot-plot of marker distribution along the P. sativum linkage groups. A flatter curve indicates a region denser in markers. The red vertical ellipses indicate gaps without markers. The blue horizontal ellipses indicate hot-spots of markers at the same genetic position

Fig. 3

Marker densification in a MetaQTL region controlling partial resistance to A.euteiches between the SSR AA505 and AB101 reference markers on PsLGVII. The left hand side shows this region on the Duarte et al. [3] consensus map and its projection on M. truncatula pseudochromosome 4 (from Duarte et al. [3] - Additional file 5: Table S4). The center shows the same region on the two individual BP-Duarte and BP-WGGBS maps, covering respectively 25 cM and 23.6 cM. The right hand side shows the same region, detailed in Hamon et al. [26], covering 52.6 cM and corresponding to three MetaQTLs Morpho8, Ae26 and Ae27 (from Hamon et al. [26] - Additional files 9: Table S6 and Additional file 10: Table S7)