Visualization of shared genomic regions and meiotic recombination in high-density SNP data

Abstract

Background: A fundamental goal of single nucleotide polymorphism (SNP) genotyping is to determine the sharing of alleles between individuals across genomic loci. Such analyses have diverse applications in defining the relatedness of individuals (including unexpected relationships in nominally unrelated individuals, or consanguinity within pedigrees), analyzing meiotic crossovers, and identifying a broad range of chromosomal anomalies such as hemizygous deletions and uniparental disomy, and analyzing population structure.

Principal findings: We present SNPduo, a command-line and web accessible tool for analyzing and visualizing the relatedness of any two individuals using identity by state. Using identity by state does not require prior knowledge of allele frequencies or pedigree information, and is more computationally tractable and is less affected by population stratification than calculating identity by descent probabilities. The web implementation visualizes shared genomic regions, and generates UCSC viewable tracks. The command-line version requires pedigree information for compatibility with existing software and determining specified relationships even though pedigrees are not required for IBS calculation, generates no visual output, is written in portable C++, and is well-suited to analyzing large datasets. We demonstrate how the SNPduo web tool identifies meiotic crossover positions in siblings, and confirm our findings by visualizing meiotic recombination in synthetic three-generation pedigrees. We applied SNPduo to 210 nominally unrelated Phase I / II HapMap samples and, consistent with previous findings, identified six undeclared pairs of related individuals. We further analyzed identity by state in 2,883 individuals from multiplex families with autism and identified a series of anomalies including related parents, an individual with mosaic loss of chromosome 18, an individual with maternal heterodisomy of chromosome 16, and unexplained replicate samples.

Conclusions: SNPduo provides the ability to explore and visualize SNP data to characterize the relatedness between individuals. It is compatible with, but distinct from, other established analysis software such as PLINK, and performs favorably in benchmarking studies for the analyses of genetic relatedness.

Figure 1. IBS patterns for father, mother, and son on chromosome X.

A portion of the SNPduo output for three pairwise comparisons of the X chromosome of father/mother (A), mother/son (B), and father/son (C) genotyped on the Illumina HumanHap 550K platform. In the unrelated parents, there were many instances of no shared alleles (e.g. AA to BB; panel A). In the mother-son comparison, there were no IBS-0 SNPs because the son inherited a copy of the maternal X. In the father/son comparison, each chromosome was hemizygous (either A or B genotypes, interpreted as AA or BB) and in the absence of heterozygous calls no IBS-1 SNPs were expected to occur since the X chromosomes were non-identical (both IBS-2 and IBS-0 SNPs were apparent). Thus, the one call of an IBS-1 SNP (arrow) was likely a genotyping error. The standard SNPduo output includes genotype calls (not shown; see e.g. Fig. 3).

Figure 2. IBS patterns for parents and a child with a hemizygous deletion.

A portion of the SNPduo output for chromosome 10 of father/mother (top panel), mother/son (middle panel), and father/son (bottom panel) comparisons genotyped on the Illumina HumanHap 550K platform. As expected the unrelated father and mother showed pattern of overlaid IBS-2, IBS-1, and IBS-0 tracks. Also as expected the mother and son comparison demonstrated a parent to child relationship of IBS-2 and IBS-1 tracks overlaid. The father to son comparison, however, demonstrated an additional pattern of IBS-2, IBS-1, and IBS-0 tracks overlaid (arrow). This block represents an interstitial hemizygous deletion of the paternal allele in the child.

Figure 3. Meiotic recombination in siblings.

Upper panel: SNPduo output for chromosome 6 comparing two siblings genotyped on the Illumina HumanHap 550K platform. The standard output format from the SNPduo web tool is shown. Three different patterns were apparent (see top panel). The region A pattern had IBS-2 SNPs alone. The region B pattern consisted of IBS-2 and IBS-1 tracks overlapping. Region C consisted of IBS-2, IBS-1, and IBS-0 tracks overlapping. IBS track types switched several times across the length of the chromosome, representing meiotic crossover sites. SNPduo output included plots of genotype calls for each of the two individuals compared on the middle and bottom panels (including summaries of the number of BB, AB, AA, and no call genotypes), the IBS state of each SNP on the top panel (including the number if IBS-2, IBS-1, and IBS-0 SNPs), and a chromosome ideogram at the bottom. Lower panel: Screen capture of SNPduo density segmentation viewed on the UCSC Genome Browser. The SNPduo output included a .bed file that was uploaded to the genome browser as a custom track. Block types corresponded to the SNPduo IBS output, as did the block type change points. Note that there are no SNP data across the centromere (see upper panel), but the segmentation crosses the centromere (note the IBS-0 block, lower panel) reflecting the absence of detectable crossovers in that region.

Figure 4. Meiotic recombination confirmed with synthetic family.

Shown are SNPduo output and calculated grandparental origin for chromosome 5 of a synthetic family (based on Illumina HumanHap 550K platform) with respect to the two siblings. Panels displayed from top to bottom: (1) SNPduo output for the siblings on chromosome 5, (2) and (3) the recombination patterns for III-1's paternal and maternal chromosome 5, and (4) and (5) the recombination patterns for III-2's paternal and maternal chromosome 5. All grandparental origin change points corresponded to IBS track type change points in the SNPduo data. Abbreviations: GM: Grandmaternal origin. GP: Grandpaternal origin.

Figure 5. Unspecified relationships in phase 1 / 2 HapMap.

Autosomal mean (x-axis) and standard deviation (y-axis) of IBS state data for the Affymetrix 500K platform visualized for all within group comparisons of a HapMap data set. Each data point corresponds to a pairwise comparison by a locally modified SNPduo. The six comparisons between related individuals are indicated as large triangles and annotated by name.

Figure 6. Inferred HapMap pedigree.

Analysis of the mitochondrial SNPs between YRI individuals NA18913 and NA19238 showed a pattern of allele sharing consistent with a parent/child relationship. The two individuals shared the same mitochondrial haplotype, suggesting that NA19238 was the mother of NA18913, making NA18913 and NA19240 half-siblings. Shaded symbols indicated shared mitochondrial heritage.