Insertion variants missing in the human reference genome are widespread among human populations

Abstract

Background Structural variants comprise diverse genomic arrangements including deletions, insertions, inversions, and translocations, which can generally be detected in humans through sequence comparison to the reference genome. Among structural variants, insertions are the least frequently identified variants, mainly due to ascertainment bias in the reference genome, lack of previous sequence knowledge, and low complexity of typical insertion sequences. Though recent developments in long-read sequencing deliver promise in annotating individual non-reference insertions, population-level catalogues on non-reference insertion variants have not been identified and the possible functional roles of these hidden variants remain elusive. Results To detect non-reference insertion variants, we developed a pipeline, InserTag, which generates non-reference contigs by local de novo assembly and then infers the full-sequence of insertion variants by tracing contigs from non-human primates and other human genome assemblies. Application of the pipeline to data from 2535 individuals of the 1000 Genomes Project helped identify 1696 non-reference insertion variants and re-classify the variants as retention of ancestral sequences or novel sequence insertions based on the ancestral state. Genotyping of the variants showed that individuals had, on average, 0.92-Mbp sequences missing from the reference genome, 92% of the variants were common (allele frequency > 5%) among human populations, and more than half of the variants were major alleles. Among human populations, African populations were the most divergent and had the most non-reference sequences, which was attributed to the greater prevalence of high-frequency insertion variants. The subsets of insertion variants were in high linkage disequilibrium with phenotype-associated SNPs and showed signals of recent continent-specific selection. Conclusions Non-reference insertion variants represent an important type of genetic variation in the human population, and our developed pipeline, InserTag, provides the frameworks for the detection and genotyping of non-reference sequences missing from human populations.

The genome atlas: navigating a new era of reference genomes

“…Over the past decade, the sharp decrease in sequencing costs has led to a flood of large-scale and population-specific sequencing projects, revealing unique sequences missing from the current reference genome [8,9], population-specific differences in common genetic variants [10,11], and increasing recognition that some populations, particularly Indigenous, are at risk of being left behind [12]. The GRCh38 reference assembly has been expanded to include alternate loci (ALT loci), most of which are similar to the primary assembly but contain many small variants that commonly occur together.…”

Population-scale detection of non-reference sequence variants using colored de Bruijn graphs

“…Others refer to NRS variants as insertions ( Delage et al , 2020 ; Wong et al , 2020 ) because the variants describe novel sequence with respect to the reference genome. However, the majority of NRS appears to be ancestral rather than novel because they can be found in other primate genomes ( Kehr et al , 2017 ; Lee et al , 2020 ). A convincing explanation for the existence of NRS is that the genomes used to construct the reference genome lacked these sequences.…”

“…Some pipelines for moderate numbers of genomes create whole-genome assemblies prior to NRS variant calling, such as the pipelines that were applied to 50 Danish trios ( Liu et al , 2015 ; Maretty et al , 2017 ), 275 Han Chinese genomes ( Duan et al , 2019 ), 1000 Swedish genomes ( Eisfeldt et al , 2020 ) and 338 genomes from genetically divergent human populations ( Wong et al , 2018 , 2020 ). Finally, pipelines developed for the 1000 genomes project data ( Lee et al , 2020 ) and for the TOPMed program ( Taliun et al , 2021 ) search for NRS variants that match related genomes like other primates’ genomes.…”

Population-scale detection of non-reference sequence variants using colored de Bruijn Graphs

“…Some pipelines for moderate numbers of genomes create whole-genome assemblies prior to NRS variant calling, such as the pipelines that were applied to 50 Danish trios [Maretty et al , 2017, Liu et al , 2015], 275 Han Chinese genomes [Duan et al , 2019], 1000 Swedish genomes [Eisfeldt et al , 2020], and 338 genomes from genetically divergent human populations [Wong et al , 2018, Wong et al , 2020]. Finally, pipelines developed for the 1000 genomes project data [Lee et al , 2020] and for the TOPMed program [Taliun et al , 2021] search for NRS variants that match related genomes like other primates’ genomes.…”

“…Others refer to NRS variants as insertions [Wong et al , 2020, Delage et al , 2020] because the variants describe novel sequence with respect to the reference genome. However, the majority of NRS appears to be ancestral rather than novel because they can be found in other primate genomes [Lee et al , 2020, Kehr et al , 2017]. A convincing explanation for the existence of NRS is that the genomes used to construct the reference genome lacked these sequences.…”