Structural variants in the barley gene pool: precision and sensitivity to detect them using short-read sequencing and their association with gene expression and phenotypic variation

Abstract

In human genetics, several studies have shown that phenotypic variation is more likely to be caused by structural variants (SV) than by single nucleotide variants (SNV). However, accurate while cost-efficient discovery of SV in complex genomes remains challenging. The objectives of our study were to (i) facilitate SV discovery studies by benchmarking SV callers and their combinations with respect to their sensitivity and precision to detect SV in the barley genome, (ii) characterize the occurrence and distribution of SV clusters in the genomes of 23 barley inbreds that are the parents of a unique resource for mapping quantitative traits, the double round robin population, (iii) quantify the association of SV clusters with transcript abundance, and (iv) evaluate the use of SV clusters for the prediction of phenotypic traits. In our computer simulations based on a sequencing coverage of 25x, a sensitivity >70% and precision >95% was observed for all combinations of SV types and SV length categories if the best combination of SV callers was used. We observed a significant (P < 0.05) association of gene-associated SV clusters with global gene-specific gene expression. Furthermore, about 9% of all SV clusters that were within 5kb of a gene were significantly (P < 0.05) associated with the gene expression of the corresponding gene. The prediction ability of SV clusters was higher compared to that of single nucleotide polymorphisms from an array across the seven studied phenotypic traits. These findings suggest the usefulness of exploiting SV information when fine mapping and cloning the causal genes underlying quantitative traits as well as the high potential of using SV clusters for the prediction of phenotypes in diverse germplasm sets.

The double round-robin population unravels the genetic architecture of grain size in barley

“…The annotation and coordinates were based on the Morex v2 gene models (Monat et al , 2019). In the next step, we used the variant calling data obtained from the whole genome sequencing project of the 23 inbreds (Weisweiler et al , 2022) to select the genes that showed polymorphisms between the parental inbreds. We examined non-synonymous (amino acid substitutions) or deleterious synonymous mutations, indels in the coding region, as well as indels and structural variants (SVs) in the 5-end regulatory region (5 kb upstream of the start codon).…”

The double round-robin population unravels the genetic architecture of grain size in barley

“…The annotation and coordinates were based on the Morex v2 gene models (Monat et al, 2019). In the next step, we used the variant calling data obtained from the whole genome sequencing project of the 23 inbreds (Weisweiler et al, 2022) to select the genes that showed polymorphisms between the parental inbreds.…”

“…We obtained SNP annotation, indels and predicted SVs information from the whole-genome sequencing of the 23 parental inbreds (Weisweiler et al, 2022). This information was used to compare the sequence polymorphism between the parental inbreds of HvDRR populations in the coding and regulatory region of the genes present in the confidence intervals of those 19 consensus QTLs that have an interval length < 10 Mbp.…”

“…We utilized the single nucleotide variant calling data, indels and predicted SVs of the 23 parental inbreds in coding and potentially regulatory regions (Weisweiler et al, 2022), and their functional annotation to describe the underlying candidate genes (Table S8, Additional File 1). We detected polymorphisms in the coding region among the contrasting parental inbreds for the genes present in the consensus QTLs.…”

“…The detailed candidate gene mining was only performed for consensus interval with a physical size of less than 10 Mb and a percent of the explained phenotypic variance of more than 10 %. Sequence polymorphism data obtained from the whole-genome sequencing project of 23 parental inbreds was used to create indels (indel below 50 bp) and predicted structural variants (indels above 50 bp, duplication and inversions) in the coding and regulatory region and SNP annotation data (Weisweiler et al ., 2022). The genes present in the consensus interval that revealed allelic differences between the contrasting groups of parental genotypes at a given QTL locus were retained as candidate genes.…”