Elevated variant density around SV breakpoints in germline lineage lends support to error-prone replication hypothesis

Abstract

Copy number variants (CNVs) are a class of structural variants that may involve complex genomic rearrangements (CGRs) and are hypothesized to have additional mutations around their breakpoints. Understanding the mechanisms underlying CNV formation is fundamental for understanding the repair and mutation mechanisms in cells, thereby shedding light on evolution, genomic disorders, cancer, and complex human traits. In this study, we used data from the 1000 Genomes Project to analyze hundreds of loci harboring heterozygous germline deletions in the subjects NA12878 and NA19240. By utilizing synthetic long-read data (longer than 2 kbp) in combination with high coverage short-read data and, in parallel, by comparing with parental genomes, we interrogated the phasing of these deletions with the flanking tens of thousands of heterozygous SNPs and indels. We found that the density of SNPs/indels flanking the breakpoints of deletions (in-phase variants) is approximately twice as high as the corresponding density for the variants on the haplotype without deletion (out-of-phase variants). This fold change was even larger for the subset of deletions with signatures of replication-based mechanism of formation. The allele frequency (AF) spectrum for deletions is enriched for rare events; and the AF spectrum for in-phase SNPs is shifted toward this deletion spectrum, thus offering evidence consistent with the concomitance of the in-phase SNPs/indels with the deletion events. These findings therefore lend support to the hypothesis that the mutational mechanisms underlying CNV formation are error prone. Our results could also be relevant for resolving mutation-rate discrepancies in human and to explain kataegis.

Figure 1.

Schematic and results of the analysis. (A) Personal real haplotypes with deletion (red) and without deletion (pure blue) are shown along with flanking SNPs/indels depicted by filled circles/rectangles. TruSeq data allows resolving SNPs/indels on the haplotype with deletion (the set of in-phase variants). The GATK variant calls (black) include SNPs/indels for both haplotypes and also provide genotype information. The reconstructed haplotype without deletion (the set of out-of-phase variants) is obtained through complement of GATK to TruSeq calls. Only heterozygous SNPs/indels are further counted. The dashed pink/green lines highlight error cases that increase the count for the haplotype with/without deletion, respectively. The pink errors are highly unlikely (see text), and the green cases can only contribute against our result. (B) Histogram showing the count of heterozygous SNPs on both haplotypes with respect to distance from the deletions’ breakpoints. A total of 262 loci with heterozygous deletions are considered. Statistically significant differences (by Z-test) in SNP counts in bins marked by star are observed in 6-kbp flanking windows. (C) Densities of heterozygous SNPs on haplotypes with and without deletions are roughly constant and higher than personal genome-wide average per haplotype. The density with deletions is twice as high as without. In-phase indel density is 33% higher than out-of-phase (not shown in figure).

Figure 2.

Densities of in-phase (red) and out-of-phase (blue) heterozygous SNPs/indels flanking deletions in (A,C) Caucasian (NA12878) and (B,D) Yoruban (NA19240) individuals. Densities are obtained from trio-based haplotype reconstruction. Average out-of-phase densities for the displayed interval are shown by dashed lines (blue). Genome-wide average densities of heterozygous SNPs per haplotype are shown by black dashed lines. In the Yoruban trio, average out-of-phase and genome-wide densities are almost the same.

Figure 3.

Histograms of normalized allele frequency distributions and normalized mutational profile for NA12878. (A) AF distribution for in-phase TruSeq SNPs (red) is between the distribution for deletions (black) and the distribution for out-of-phase TruSeq SNPs (blue), suggesting that it is a superposition of the two. This pattern is consistent with the hypothesis of SNPs being generated simultaneously with the deletions. (B) Mutational profile for in-phase (red) and out-of-phase (blue) TruSeq SNPs. The six possible transversions and transitions are shown, with the normalized count on the vertical axis. The third bar (black) represents personal genome-wide profile based on heterozygous SNPs.