Single‐nucleotide polymorphism discovery and validation in high‐density SNP array for genetic analysis in European white oaks

Abstract

An Illumina Infinium SNP genotyping array was constructed for European white oaks. Six individuals of Quercus petraea and Q. robur were considered for SNP discovery using both previously obtained Sanger sequences across 676 gene regions (1371 in vitro SNPs) and Roche 454 technology sequences from 5112 contigs (6542 putative in silico SNPs). The 7913 SNPs were genotyped across the six parental individuals, full-sib progenies (one within each species and two interspecific crosses between Q. petraea and Q. robur) and three natural populations from south-western France that included two additional interfertile white oak species (Q. pubescens and Q. pyrenaica). The genotyping success rate in mapping populations was 80.4% overall and 72.4% for polymorphic SNPs. In natural populations, these figures were lower (54.8% and 51.9%, respectively). Illumina genotype clusters with compression (shift of clusters on the normalized x-axis) were detected in ~25% of the successfully genotyped SNPs and may be due to the presence of paralogues. Compressed clusters were significantly more frequent for SNPs showing a priori incorrect Illumina genotypes, suggesting that they should be considered with caution or discarded. Altogether, these results show a high experimental error rate for the Infinium array (between 15% and 20% of SNPs potentially unreliable and 10% when excluding all compressed clusters), and recommendations are proposed when applying this type of high-throughput technique. Finally, results on diversity levels and shared polymorphisms across targeted white oaks and more distant species of the Quercus genus are discussed, and perspectives for future comparative studies are proposed.

The loops of the N-SH2 binding cleft do not serve as allosteric switch in SHP2 activation

“…The correlations among maf in both datasets were high and significant (0.66 and 0.68 respectively for Q. petraea and Q. robur, both Pr< 0.0001). These results suggest a small risk of SNP ascertainment bias if the new resources were to be used in populations within the geographic distribution surveyed, in contrast to panels with much less individuals (see Lepoittevin et al 2015 for a discussion on the consequences of such bias in Quercus species).…”

“…We therefore do not recommend the development of a very large SNP array in oaks since it is likely to be very costly for a minimal return, especially given the very large and range-wide panel that would be needed to significantly limit ascertainment bias (see Lepoittevin et al 2015). Also the very high SNP density required for targeting of functional variants given low overall levels of LD (indicating potentially high recombination rates), would be technically constrained when controlling for genotyping error rates (shown previously to be high).…”

“…Recently, building from the European oaks genomic resources (Quercus Portal, htpps://arachne.pierroton.inra.fr, and references therein), natural populations of 4 Quercus species (Q. robur, Q. petraea, Q. pyrenaica, Q. pubescens) were genotyped for ~4000 singlenucleotide polymorphisms (SNPs, from an initial 8K infinium array, Lepoittevin et al 2015).…”

“…Indeed, very stringent filters are often applied in practice to limit error rates and avoid false-positives, hence limiting the impact of variable read depth and possible ascertainment bias risks, which altogether significantly decrease the number of informative loci compared to either initial fixed amounts (in genotyping arrays, e.g. Lepoittevin et al 2015) or potential amounts (in reference genomes, e.g. Pina-Martins et al 2018 in Quercus species; see also Van Dijk et al 2014).…”

“…http://dx.doi.org/10.1101/388447 doi: bioRxiv preprint first posted online Aug. 10, 2018; another one (Qs28) from central France, considered previously to be a Q. petraea parental genotype in two oak mapping pedigrees (Bodénès et al 2012(Bodénès et al , 2016Lepoittevin et al 2015). …”