GWAS SVatalog: a visualization tool to aid fine-mapping of GWAS loci with structural variations

Genome-wide association studies (GWAS) have been successful in identifying genetic variants associated with human traits (Visscher et al., 2017; Uffelmann et al. 2021). GWAS identification of a protein-coding variant informs understanding of the mechanism by which an associated variant contributes to the analyzed trait, however, 93% of GWAS associated loci (GAL) identified lie in non-coding regions of the genome (Maurano et al., 2012). Such regions frequently harbor regulatory features but the mechanism by which the associated locus impacts the trait is not immediately obvious. These features may be in proximity to the affected gene or farther away where the 3D structure of the chromosome contributes to regulation (Orozco et al., 2022). GWAS typically use single nucleotide polymorphisms (SNPs) (Visscher et al., 2017; Uffelmann et al., 2021), which are straightforward to study for association on a large scale as they are mostly biallelic and highly quantifiable using microarray and/or short-read whole-genome sequencing (WGS). However, SNPs account for an incomplete proportion of genetic variation and phenotypic heritability (Zarrei et al., 2015), and it is unknown how often they tag other polymorphisms, such as structural variants, that may be the cause of the association signal.

A structural variant (SV) is a large genetic polymorphism, typically defined to range from 50 bp to several megabases and includes insertions, deletions, duplications and inversions. SVs can have a pronounced effect on gene regulation with downstream phenotypic consequences (Chiang et al., 2017; Vialle et al., 2022; Trost et al., 2022). Due to their size and complexity, identification and genotyping of SVs is difficult using short-read sequencing technologies. Short reads often fail to map correctly near a large SV due to reference bias (Meynert et al., 2014). Due to the size of short reads, most SVs are not found within a single read; their identification requires overlapping short reads to infer an SV signature (Mahmoud et al., 2019). Databases that catalog SVs from technologies that include short reads exist, such as the Database for Genomic Variants (DGV) (MacDonald et al., 2014) and the Genome Aggregation Database (gnomAD) (Chen et al., 2022). These are remarkable public resources used extensively by researchers, but the content of these resources is limited by the technology that is used. Since the advancement of long-read sequencing technologies, the accuracy in detecting complex SVs has increased dramatically due to longer read lengths, improved de novo assemblies and new SV-specific callers (PacificBiosciences, 2018). However, despite these advancements, SV detection, characterization, boundaries and allele-differentiation remains imperfect. Nevertheless, we can substantially augment these resources to aid in fine-mapping of GAL by calling common SVs from long-read WGS and pre-computing their linkage disequilibrium (LD) with GWAS SNPs, which we demonstrate here with our new tool, GWAS SVatalog.

SVs are important to genotype accurately as they have been previously shown by fine-mapping to explain GWAS signals. A motivating example from our own work is a GWAS of intestinal obstruction at birth in cystic fibrosis (CF), where we identified a suggestive locus at chr7q35 (Gong et al., 2019). Fine-mapping of this region using linked-read WGS uncovered a large 20-kb deletion SV in high LD with the GWAS-suggestive SNPs (Mastromatteo et al., 2023). We show that this SV is an eQTL for serine protease 2 (PRSS2) which encodes the digestive enzyme anionic trypsinogen (Mastromatteo et al., 2023), and is the putative cause for the GWAS locus. Discovery of SVs as the cause of GAL has been reported by others as well (Fritsche et al., 2008; Trost et al., 2022). For example, in a GWAS investigating age-related macular degeneration (AMD), a study first identified rs10490924 as associated with disease risk (Rivera et al., 2005) and this was replicated in multiple independent studies (Jakobsdottir et al., 2005; Fritsche et al., 2008; Micklisch et al., 2017). Fine-mapping of the GWAS locus identified an SV in high LD with rs10490924 located in the 3’UTR of ARMS2 (Fritsche et al., 2008). This SV was shown to remove the poly-A tail of the transcript and affect expression of the protein (Micklisch et al., 2017), ultimately influencing the risk of AMD.

Many GWASs have identified trait-associated SNPs with no reported cause (Chen et al., 2021; Stachowska et al., 2022; Lee et al., 2023). Here we create a population level catalog of common SVs called from long-read WGS. The SV calls were derived from sequencing the DNA of 101 individuals enrolled in the CF Canada-SickKids Program in Individualized Therapy (CFIT) (Eckford et al., 2019), using both PacBio continuous long-read (CLR) and 10X Genomics (10XG) linked-read technologies. Using two separate sequencing platforms is rare for SV catalogs and has the benefit of increasing the reliability and robustness of the calls. Current publicly available long-read databases are not population-based and contain diverse ethnicities with small sample sizes (Liao et al., 2023; Logsdon et al., 2024), minimizing the reliability of LD required for this study’s analysis. We show here that the allele frequencies (AF) across the genome of individuals with CF are comparable to individuals of European origin in the general population. The curated list of SVs cataloged here has been benchmarked against well-studied genomes and annotated variant databases. Using our sequences, we compute the LD between 35,732 SVs and 116,870 GWAS-associated SNPs provided in the GWAS Catalog (Sollis et al., 2023), thereby evaluating association between common SVs and 14,479 human traits. We aid fine-mapping and functional follow-up for the research community by cataloguing and visualizing the LD in our novel web tool, GWAS SVatalog (https://svatalog.research.sickkids.ca/), as part of the LocusFocus (Panjwani et al., 2020) suite of software tools.

Sample cohort and DNA extraction

CFIT is a collaborative project and biospecimen repository of participants with CF to aid in advancing personalized CF therapies (Eckford et al., 2019). CFIT recruited and obtained whole blood from 101 Canadians with CF. Methods for recruitment, biospecimen collection and data generation are described in Eckford et al. (2019). Here we use the resulting sequencing data.

Library preparation and sequencing

PacBio CLR

Library preparation and DNA extraction details are provided in Eckford et al. (2019). PacBio Sequel I and II were used to carry out the long-read sequencing on 34 and 67 samples, respectively. Although the samples were sequenced on two different platforms, we did not find a significant difference between their average read lengths. The average coverage for both machines is 50x and 76x respectively.

10XG Linked-Reads

Library preparation and DNA extractions details are provided in Eckford et al. (2019). Illumina HiSeq X was used to carry out the paired-end sequencing of approximately 150 bases in length at 30x coverage.

Variant calling

For each sample, we used a combination of callers for each sequencing platform previously shown to achieve the best results (Coutelier et al. 2022; Mahmoud et al. 2024). To align and call SVs from the PacBio CLR sequences, the combination of pbmm2 1.1.0 (PacificBiosciences, 2020) + pbsv 2.2.2 (PacificBiosciences, 2018) and NGMLR 0.2.8 (Sedlazeck et al., 2018) + Sniffles 1.0.11 (Sedlazeck et al., 2018) was used. For the 10XG sequences, Long Ranger 2.2.2 (10XGenomics, 2020) was used for sequence alignment, and SV calling was conducted on Long Ranger, CNVnator 0.4 (Abyzov et al., 2011), ERDs 1.1 (Zhu et al., 2012) and Manta 1.6.0 (Chen et al., 2016). All software was run using GRCh38 including its alternative contigs as the reference genome at default settings except for the –min_support parameter on Sniffles, which was set to five minimum reads. Samples run on Sequel I had their minimum read threshold reduced to five reads, instead of the default of ten due to lower coverage. SNPs were called from the 10XG sequences in Long Ranger using the Haplotype Caller function in GATK 4.0.0.0 (Van der Auwera and O’Connor, 2020) with default parameters.

Comparison of allele frequencies between the CF cohort and EUR population from 1000 Genomes Phase 3

Using the list of high-confidence SNPs taken from Illumina Omni 2.5-8 v1.5 (https://webdata.illumina.com/downloads/productfiles/humanomni25/v1-5/infinium-omni2-5-8v1-5-a1-manifest-file-csv.zip), we compared the AFs of these SNPs between our CF cohort and the EUR population from 1000 Genomes Phase 3 (1000 Genomes Project Consortium et al. (2015)) using Fisher’s Exact Test in PLINK 1.90beta3a (Purcell et al., 2007). Aside from the CF-causing variants on chromosome 7 and the LD block in which they lie, the remainder of the genome does not differ significantly in AF from the 1000 Genomes European population (Supplementary Fig. 1).

Merging SV calls

For the PacBio CLR sequences, pbsv and Sniffles were utilized while the 10XG sequences used Long Ranger, CNVnator, ERDs, and Manta. SV calls were finalized after three merging steps: within the PacBio CLR sequences, within the 10XG sequences, and between the two sequencing platforms (PacBio CLR and 10XG). In all steps, a 50% reciprocal overlap rule was used for deletions, inversions and duplications where the SV boundaries are finalized as the average breakpoints from the constituting calls. For insertions the criteria were slightly different, where we made sure the breakpoints were within 1,000 bp and the length of the SVs within 50% of one another. The rendered call may not have the exact coordinates of the polymorphism itself but is representative of the region indicating a prominent SV.

The first step commenced by filtering pbsv calls with similar breakpoints in repeat regions. This was done to avoid duplication but does not take into account varying alleles in the individual. The pbsv calls are used as an anchor while the calls made by Sniffles were used as supporting evidence. When both pbsv and Sniffles identify the same SV, the Sniffles call would “tag” the pbsv coordinates and only the pbsv call (and genotype) would be used moving forward. This process of tagging is used to help eliminate repetition. As pbsv uses consensus sequences from supporting reads to determine SV boundaries, we chose to use its output as the primary result. However, all non-tagged calls (unique calls) made by each software are also retained moving forward. In the second step, Manta calls were used as the anchor and the supporting calls were sequentially added in the order of CNVnator, ERDs followed by Long Ranger. Only deletion calls were used from Long Ranger as the software does not identify insertions and the calls made for duplications and inversions were disproportionate in comparison to the other software. The last merging step follows the technique used in Audano et al. (2019) for merging SV calls across both sequencing platforms. The boundaries defining the SV are the ones from the PacBio CLR calls when present in both sequencing platforms.

To complete the merging step, SVs across all samples were consolidated to create a comprehensive SV database. SVs with overlapping coordinates were only included into the final data set if at least one call was greater than 50 bp in length. Each merging step was carried out using bcftools merge 1.20 (Danecek et al., 2021).

Lastly, SV calls which appear in fewer than three participants were removed from the database for the purpose of maintaining GA4GH principles (Rehm et al., 2021), bringing the total SVs in the database down from 129,485 to 87,183 SVs. The overview of the pipeline implemented can be seen in Fig. 1.

Reliability of SV calls

The SV database was validated based on SV boundaries and non-reference allele frequency (NAF). Truvari (English et al. 2022) was used to compare SV calls with three public SV datasets (Zook et al. 2016; Audano et al. 2019; Beyter et al. 2021) using the parameters truvari bench –refdist 500 –pctsize 0.7 –pctoval 0.5. The SV dataset by gnomAD v4.1 (Chen et al. 2024) was used to compare SV boundaries and NAF to this study. We calculated the concordance correlation coefficient (CCC) in the two datasets and reported the confidence intervals (CI). In addition, our SV genotypes were validated by using HG002 sequences for both 10XG and PacBio CLR from Genome In A Bottle (GIAB) (https://github.com/genome-in-a-bottle/giab_data_indexes). These sequences were put through the same SV calling pipeline as our samples and the genotypes were compared with the draft release HG002 benchmark file from GIAB (https://ftp-trace.ncbi.nlm.nih.gov/ReferenceSamples/giab/data/AshkenazimTrio/analysis/NIST_HG002_DraftBenchmark_defrabbV0.019-20241113/GRCh38_HG2-T2TQ100-V1.1_stvar.vcf.gz). The comparison was carried out using the same Truvari parameters outlined above which provides genotype matching statistics.

Linkage disequilibrium statistics

LD refers to the nondependent inheritance of alleles where, in this study, we focus on the relationship between SNPs and SVs. The two statistics, r² and D’, are calculated and indexed in the tool to capture LD between the polymorphisms. In order to begin calculating the LD statistics, Picard MergeVcfs 2.26.8 (Broad Institute, 2019.”) is used to consolidate the SV calls from both sequencing platforms and SNP calls from the 10XG sequences. The SNP calls in the 101 samples are filtered by selecting the SNPs reported in GWAS Catalog with a p-value < 10⁻⁶. Next, SHAPEIT4.2.0 (Delaneau et al., 2019) is used to estimate phase on the variant calls as the SNP calls from PacBio CLR sequences are presumed to be error prone. Phasing error rates using SHAPEIT4 have been shown by previous researchers to be low for both PacBio (0.23%) and 10XG (0.07%) (Delaneau et al., 2019). LD statistics are calculated for GWAS Catalog SNPs within 1 Mb of each SV by PLINK 1.90b3x.

Annotation of SNPs and SVs

The custom annotation pipelines for SNPs and SVs were developed at The Centre for Applied Genomics (TCAG) at The Hospital for Sick Children and have been implemented in several studies (Chan et al., 2022; Trost et al., 2022). The SV calls were annotated using a custom CNV/SV annotation pipeline based on UCSC tracks (Nassar et al., 2023), gnomAD v4.1, Decipher v11.25, ClinVar/ClinGen (Landrum et al., 2020) portals and DGV gold standards (MacDonald et al., 2014). The SNPs were likewise annotated using a custom small variant annotation pipeline based on ANNOVAR v2019Oct24 (Wang et al., 2010), gnomAD v3.1 (Gudmundsson et al., 2022) and dbSNP v138 (Sherry et al., 2001).

Merging with public data

Merging our LD statistics with GWAS Catalog uses the genomic position of SNPs, their reference and alternate allele. Supplementary Fig. 2 shows the details of each merging step and how the data was incorporated into the database used by GWAS SVatalog. GWAS Catalog v1.0-associations_e108 was obtained from https://www.ebi.ac.uk/gwas/docs/file-downloads. MANE Select (Morales et al., 2022) from https://ftp.ncbi.nlm.nih.gov/refseq/MANE/MANE_human/release_1.0/MANE.GRCh38.v1.0.ensembl_genomic.gtf.gz was used to visualize genomic location in GWAS SVatalog. The curated SV database along with SNP calls and their LD statistics (r² and D’) data is publicly accessible in a Zenodo repository at https://zenodo.org/records/13367574. The database indexes 35,732 SVs associated with phenotypes from the GWAS Catalog.

GWAS SVatalog: web application development

GWAS SVatalog (https://svatalog.research.sickkids.ca/) is a web-based visualization tool built using Plotly Dash and Python (Fig. 2). It is part of the LocusFocus suite of software tools (https://locusfocus.research.sickkids.ca/) (Panjwani et al., 2020) specializing in data integration for GWAS follow-up. GWAS SVatalog allows users to visualize the LD between 35,732 common SVs and 116,870 GAL found for 14,479 phenotypes from the GWAS Catalog. For a detailed description on how to use the tool, visit the GWAS SVatalog’s documentation (https://gwas-svatalog-docs.readthedocs.io/en/latest/index.html).

SV calls

We devised a pipeline (Fig. 1) that merges SV calls made by pbsv and Sniffles from our PacBio CLR long-read sequencing data, with SV calls from Long Ranger, CNVnator, ERDs, and Manta from our 10XG linked-read sequencing data. In total, 129,485 distinct SVs >50 bp were identified with an average size of 977 bp. The N50 read length range of 22–35 kb from the long-reads resulted in most SVs detected within a single read. We observe 60,591 insertions and 63,301 deletions, which is higher than previously reported genome-wide long-read SV call sets (Audano et al., 2019; Beyter et al., 2021).

From the total 129,485 SV calls, 87,183 distinct SVs were found in at least three individuals from our sample cohort. Each of these SVs have been annotated with their basic information such as length, frequency and type, in addition to more detailed annotations such as percentage of GC content, genes they overlap, repeat region overlap, and percentage overlap with gnomAD populations.

Generalizability and validation of SV calls

Our SVs were called from the DNA of 101 individuals with CF who are predominately of European origin (Eckford et al., 2019) however, these SV calls and frequencies are generalizable to non-CF individuals of European origin. This is because, although the SVs were called from individuals with two CF-causing mutations in the CF transmembrane conductance regulator (CFTR) on chromosome 7, we show the genetic background of our cohort does not differ significantly from that of a healthy European population, with the exception of the LD block encompassing CFTR.

For this analysis, we compared the genotypes of high-confidence SNPs from the CF cohort to the EUR population of 1000 Genomes Phase 3 (1000 Genomes Project Consortium et al., (2015)) (Supplementary Fig. 1) demonstrating no significant differences in AFs outside of the CFTR locus using Fisher’s exact test. In our cohort of 101 individuals, 91 are of European origin while the remaining 10 are from varying demographics including African and Asian. The estimation of ancestry of the cohort was conducted using GrafPop (Jin et al. 2019). The most common CF-causing mutation is F508del, with 51 homozygous carriers and 20 heterozygotes. Additional cohort demographics are provided in Supplementary Table 1.

We compared our SV calls to three publicly available SV sources: HG002 (Zook et al., 2016); a 15 multi-ethnic sample cohort using PacBio long-reads (Audano et al., 2019); and SVs called in 3,622 Icelandic individuals using Oxford Nanopore long-reads (Beyter et al., 2021). The comparison with our SV database was carried out using Truvari based on the NAF of the polymorphism. NAF is the percentage of alleles that deviate from the reference at this SV site. As allelic differentiation has not been made in this dataset, any deviation from the reference genome, GRCh38, in the stated boundary classifies as an SV. For example, an SV in our cohort with an NAF of 0.57 indicates that 57% of the haplotypes in our cohort deviate from GRCh38 at the location of this SV. Of the 8,902 SVs in our population with NAF > 0.5, 94% of these SVs were also called in the three publicly available SV sources we investigated. We then looked at 31,087 SVs with NAF > 0.1 in our population and found 85% of these SVs were also seen in the three comparison sources.

To compare SV genotype calls, we used publicly available HG002 sequences from both 10XG and PacBio CLR, and the HG002 draft benchmark file from GIAB. After running the sequences through our SV calling pipeline, we found an 82% concordance rate using Truvari bench.

When comparing our SV catalog to gnomAD v4.1, a recently updated public database created using short-read sequences, we see a 53.46% overlap. Here we find that a lower percentage of our calls match the SVs in this database (Supplementary Table 2). Similar to the long-read dataset comparison, we subsetted SVs based on the NAF of the polymorphism. From the SVs in gnomAD we identify 68% and 72%, respectively for SVs in our cohort with NAF > 0.1 and NAF > 0.5. This aligns with the assumption that short-read data is limited in the SVs it can genotype. Additionally, from the overlap of SVs, we compared NAF values by calculating the CCC between the two datasets, resulting in 0.736 (95% CI: 0.720–0.749).

Distribution of SVs across the genome

The distribution of SV lengths shows prominent peaks at the 300 bp and 6,000 bp lengths corresponding to Alu and LINE elements, respectively (Collins et al., 2020) (Supplementary Fig. 3). A high density of SV calls can be seen in telomeric regions (Supplementary Fig. 4) with most of the singleton calls detected there. The number of SVs overlapping regulatory features, CpG islands, repeats, segmental duplications, and topologically associated domain (TAD) blocks can be found in Supplementary Tables 3, 4. We see that enhancers are frequently overlapping with SVs. This is intriguing as enhancers affect gene expression within a given TAD block (Panigrahi and O’Malley, 2021) and the TAD block boundaries demarcate the 3D conformation of a locus (McArthur and Capra, 2021). There are 73,655 unique common SVs that are present within TAD blocks, and we observed only 72 SVs to overlap TAD boundaries, which can have functional consequences when altered such as disruption of gene expression (Panigrahi and O’Malley, 2021).

Supplementary Table 4 provides the number of unique SVs that overlap with gene boundaries (i.e. start and/or stop codons) potentially affecting transcription of the gene. The table also includes 1,062 unique SVs overlapping with exon/intron boundaries with the potential of interrupting splice sites and creating non-functional transcripts. We have also noted a small percentage of SVs overlapping entire genes as shown in Supplementary Table 5. As seen in the table, there is one deletion SV encompassing 21 genes. This SV is depicted as a common SV in other publicly available databases such as dbVar (Lappalainen et al., 2013) and DGV (MacDonald et al., 2014). The DUB/USP17 gene family is among the 21 genes deleted by this SV. They are a highly conserved family of genes within and among mammalian species consisting of a high proportion of tandem repeats (Burrows et al., 2005; Yang et al., 2021).

Association testing using logistic regression (Methods detailed in Supplementary materials) indicated that higher SV NAF was associated with lower odds of CpG Island overlap (p < 0.0001) and lower odds of Promoter overlap (p = 0.0123). SV Type was significantly associated with regulatory feature overlap, although the most/least likely SV types to overlap regulatory features varied between regulatory features (p ranging from <0.0001 to 0.0377). Higher SV size and greater numbers of GWAS SNPs near an SV led to higher odds of SV overlap with each regulatory feature (all p < 0.01). SVs that were less directly tagged by GWAS SNPs (based on either D’ or r²) tended to overlap CpG islands (p < 0.0001) and promoters (all p ≤ 0.0445). A summary of SV length, type and NAF is provided in Supplementary Table 6.

Identifying GWAS loci that may be explained by an SV using GWAS SVatalog

GWAS SVatalog is capable of queries based on a phenotype of interest or a genomic region of interest in GRCh38 coordinates. Figure 3 provides examples of analyzing a single target SV. Figure 3A provides sample output of LD between a chosen target SV and GWAS-significant SNPs for all phenotypes while Fig. 3B shows an alternate view of the target SV, highlighting the LD with GWAS-significant SNPs of a specified target phenotype. To conduct further investigation, all data in the plot, including phenotype associations, can be downloaded by the user.

An overview of the distribution of SVs available in GWAS SVatalog based on their type (deletion, duplication, insertion and inversion) and annotation by functional genetic regions can be found in Table 1. Supplementary Table 7 provides the proportion of SVs associated with GAL using the max LD score per SV. There are 9,438 SVs, located within a gene, in high LD (D’ ≥ 0.8) with GAL, with 530 GWAS-significant SNPs being exonic. Of the 36,295 SVs with MAF ≥ 0.1, 35,732 SVs had SNPs within 1 Mb of their boundaries with non-zero LD. In total, there are 116,870 GAL. Among these, 64,919 GAL are in high LD (D’ ≥ 0.9) with 21,578 SVs.

Using the GWAS SVatalog and the SV calls that we generated, we wanted to identify SVs that may explain GAL. We first filtered 21,578 SVs in high LD (D’ ≥ 0.9) with GAL from where we only considered 12,002 SVs that overlap protein-coding genes. Of these, we focused on 9,914 SVs in high LD with two or more phenotypes (i.e. pleiotropic/replication). We then randomly selected 100 SVs from this subset and visualized each SV using GWAS SVatalog. We identified the previously reported causal deletion SV present in the 3’UTR of ARMS2 associated with age-related macular degeneration (Fritsche et al., 2008; Micklisch et al., 2017) (Fig. 4A). We also identified three novel candidates, shown in Table 2 and Fig. 4B–D, where an SV appears to potentially affect regulatory activity which could explain the GWAS signals.

The first candidate SV (Fig. 4B), a 1317 bp deletion (NAF = 0.342) localized at the 3’UTR of TF on chromosome 3 was found to be in LD (D’ = 0.952) with rs3811647 (AF = 0. 366), a GWAS-significant SNP in an iron biomarker GWAS (Benyamin et al., 2009). This SV is a SINE-VNTR-Alu (SVA) retrotransposon that removes a SREBP2 binding site. The second candidate (Fig. 4C), a 54 bp deletion (NAF = 0.114) in intron 1 of KCNQ5 on chromosome 6 was in high LD (D’ = 1) with rs7744813 (AF = 0.614), associated with refractive error (Verhoeven et al., 2013). This intronic SV has the potential to influence KCNQ5 expression as it is located within H3K4me1 and H3K27ac histone marks and two transcription factor ChIP-seq clusters (ZNF263 and PCBP2) known to be transcriptional and translational suppressors, respectively (Ren et al., 2016; Yu et al., 2020). The third candidate (Fig. 4D), a 323 bp deletion localized to the 3’UTR of TMEM106B on chr7 (NAF = 0.406) is in perfect LD (D’ = 1) with rs1990622 (AF = 0.406), associated with Alzheimer’s disease (Hu et al., 2021). This SV is a SINE element that removes multiple epigenetic elements: three histone marks (H3K4Me1, H3K4Me3, and H3K27Ac), a CpG island, and two transcription factor binding sites (TBX21 and PKNOX1).

In these examples, the GAL have no known functional consequence tying phenotype to genotype, to the best of our knowledge. GWAS SVatalog provided an alternative explanation for these GAL by visualizing SVs in high LD that could be putatively functional. These are but a small subset of examples identified from a large list of candidate SVs. Further investigation using GWAS SVatalog has the potential to identify many other SVs that could explain GALs and lead to translational discoveries.

Here we introduce a novel web tool, GWAS SVatalog, that integrates SVs into visualizations of GAL to assist in fine-mapping and putative causal variation identification. We created a catalog of SVs from a cohort of individuals representative of populations of European origin using a combination of PacBio CLR sequencing and 10XG linked-read sequencing. By utilizing both of these technologies, we are able to leverage their best attributes: CLR can identify large polymorphisms, like SVs, and 10XG excels in genotyping SNPs. The SV call set we produced, and corresponding LD calculations with respect to SNPs reported in the GWAS Catalog, has been made available through the GWAS SVatalog visualization tool. The web tool aids fine-mapping by incorporating 35,732 SVs at GAL and visualizing the LD relationship between these SVs and 116,870 GWAS-significant SNPs from 14,479 different human traits.

For our cohort, we demonstrated that the AFs across the genome (with the exception of the CFTR locus) are representative of the EUR sub-group of the 1000 Genomes healthy control group (Supplementary Fig. 1). This gives us confidence in the ability of GWAS SVatalog to represent accurate LD relationships for a GWAS comprised of a high proportion of Europeans. For other ethnicities, this remains a major limitation of the tool which we are hoping to address by including more diverse SV call sets as they are generated, such as those included in the All of Us study (All of Us Research Program Genomics Investigators, 2024).

In 2020, gnomAD released a dataset linking common SVs genotyped from short-read sequences with GWAS variants (Collins et al., 2020). However, SV identification is limited when genotyped using short-read sequence data due to their average read length of 150 bp preventing larger SVs that can span thousands of base pairs to be called efficiently (Mahmoud et al., 2019). Short-reads also have difficulty calling SVs that lie in repeat regions (Collins et al., 2020; Kosugi and Terao, 2024), are related to copy number variations (Collins et al., 2020), or have a combination of events (complex) such as inversion-duplication (Sedlazeck et al., 2018; Collins et al., 2020). Overall, short-read sequences have been shown to miss about 30% of SV calls compared to long-read sequences (Sedlazeck et al., 2018). Supplementary Table 2 shows that our data aligns with this estimate as well when comparing the newest short-read gnomAD SV database to our long-read SV database.

Benchmarking conducted on our SV database against three publicly available long-read SV resources (Zook et al., 2016; Audano et al., 2019; Beyter et al., 2021) supports the reliability of our SV calls as 85% of the common SVs (NAF > 0.1) were found in the other three datasets. The three sources were derived from varying sequencing platforms and software calling methods, contributing to the differences in the SVs identified.

Association testing showed that SV NAF was inversely associated with overlap of CpG islands and promoters, which is consistent with existing literature (Schloissnig et al. 2025). Both CpG island overlap and promoter overlap were also associated with SV-SNP LD. Specifically, SVs that are less directly tagged by GWAS SNPs (whether measured by maximum D’ or r²) were more likely to overlap CpG islands or promoters than to overlap no regulatory feature, after controlling for SV size, type, and NAF. This is consistent with the hypothesis that some GWAS hits may be driven by SVs altering the function of nearby genes.

We used GWAS SVatalog to identify three candidate SVs that may explain the GWAS signal reported in the literature, since the functional polymorphism was not identified to our knowledge. These SVs lie in regions within genes and/or are overlapping regulatory features that could affect the level of gene expression. For example, the 54 bp deletion SV in the first intron of KCNQ5 is in high LD (D’ = 1) with an intronic variant, rs7744813. This SNP has not been shown to be a causal variant but is repeatedly shown to be significantly associated in GWASs of refractive error and myopia (Verhoeven et al. 2013; Li et al. 2021; Liao et al. 2017). As this SV removes part of two transcription factor binding sites, ZNF263 and PCBP2, which have previously been shown to be transcriptional and translational suppressors (Ren et al., 2016; Yu et al., 2020), further analysis of their interaction with KCNQ5 could potentially demonstrate the significance of the GWAS signal at this locus. The second candidate SV identified using GWAS SVatalog is a 323 bp deletion in the 3’UTR of the gene TMEM106B. This SV is in complete LD (D’ = 1 and r² = 1) with a variant, rs1990622, located downstream of the gene. Even though rs1990622 has been shown to be highly associated in GWAS of Alzheimer’s disease (Hu et al., 2021), it is also known to impact other neurological disorders such as frontotemporal dementia and Parkinson’s disease (Van Deerlin et al., 2010; Lee et al., 2023). Studies have linked rs1990622-A to an increase in levels of TMEM106B protein in the lysosome leading to an increased risk of dementia (Lee et al., 2023). As this SNP and the SV are in complete LD, functional studies of this SV could determine whether the SV could affect the neurological phenotypes. For the third candidate SV, a 1317 bp deletion is seen in high LD (D’ > 0.95) with an intronic variant rs3811647 that has been repeatedly associated with a GWAS of variation of serum transferrin levels (Benyamin et al., 2009; Pichler et al., 2011; Blanco-Rojo et al., 2012). Transferrin is essential for the homeostasis of iron levels by aiding in its transportation to cells (Blanco-Rojo et al., 2012) and the role of rs3811647 in the expression of the transferrin gene (TF) has not yet been reported. As this SV is a SINE-VNTR-Alu (SVA) retrotransposon and encompasses a few transcription factor binding sites (including SREBP2), further functional analyses will determine if this SV impacts TF expression and variation of serum transferrin. As supporting evidence, SVAs have been reported to have a pronounced effect on gene expression (Hancks and Kazazian, 2010; Quinn and Bubb, 2014) and SREBP2 has been shown to directly affect the transcription of TF in circulating tumor cells (Hong et al., 2021). Therefore, upon functional follow-up, this SV could explain the significance of the GWAS signal. All three candidate SV examples illustrate how GWAS SVatalog can improve fine-mapping of a locus without genotyping of the SVs in the original GWAS discovery sample and can direct future studies for causal variant identification.

Although GWAS SVatalog provides a unique contribution to fine-mapping moving forward, we note several limitations. First, the data produced for GWAS SVatalog uses SV calls from reference-aligned long-reads which improves SV identification in contrast to short reads but is still dependent on alignments that are susceptible to reference bias. Reference bias is a systematic error that occurs during the alignment of sequencing reads to a reference genome resulting in underrepresentation of variation that differs significantly from the reference genome. It produces errors such as false-negatives, false-positives, and incomplete detection of complex SVs (Günther and Nettelblad, 2019; Martiniano et al., 2020; Valiente-Mullor et al., 2021). More specifically, the development of the GWAS SVatalog tool was motivated by a 20 kb deletion polymorphism tagged by a suggestive SNP in a GWAS of intestinal obstruction in CF (Gong et al., 2019; Mastromatteo et al., 2023). This finding led us to hypothesize that similar cases might exist in other contexts, where GWAS-significant SNPs are not the variation that explain the GWAS signal but rather markers of larger, functionally significant variations. Despite being the motivation of this work, we were unable to accurately call the 20 kb polymorphism using PacBio CLR data aligned to GRCh38. Even with PacBio HiFi data (data not shown here), we were still unable to call the 20 kb insertion against GRCh38 despite the increase in base pair accuracy. Therefore, the SVs cataloged here are an incomplete subset of all SVs in our population. Future work with de novo assemblies and reference pangenomes can potentially improve SV calling by mitigating the effect of reference bias.

The second limitation arises from the SV calling software currently available. We were unable to fully characterize the alleles of each SV due to difficulty in calling their boundaries. Due to this challenge, we collapsed all SVs in the same position across samples into one non-reference SV allele, leading us to treat all SVs in this dataset as biallelic. As a result of collapsing non-reference alleles together, the reported NAF of the SVs does not fully capture the nuance of each allele. As some SVs will be multiallelic, we are unable to showcase that variation for now. However, even if the SVs were captured to be multiallelic, there is no standard methodology to calculate LD for a pair of loci belonging to biallelic and multiallelic variants, bringing us to the third limitation of all polymorphisms in this dataset being treated as biallelic. Due to this, the resulting LD between GWAS-significant SNPs and SVs may be slightly overestimated (Hedrick, 1987; Gaunt et al., 2006; Zhao et al., 2007; Jiang et al., 2020). However, the D’ coefficient is a reliable statistic with respect to the allelic discrepancy as its observation has been shown to be predominately frequency independent (Hedrick, 1987; Zapata, 2000; Zhao et al., 2007). As a significant proportion of SV-SNP pairs have low r² and high D’, SNPs seem to be tagging SVs with differing frequencies from SNPs, possibly reflecting the multi-allelic nature of SVs.

The integration of SVs in GAL will fill the gap in knowledge created by primarily using SNPs in GWAS. Using GWAS SVatalog to guide functional follow-up by visualizing LD between common SVs and GALs will aid in fine-mapping to better understand the mechanism of action at the GAL. We are confident in the curation of our SV database built primarily from long-reads and hope to make future improvements by using HiFi sequences and pangenome references to alleviate the challenges imposed by SV genotyping and reference bias. In the future we hope to expand the long-read sequencing data used to call the SVs that are cataloged in GWAS SVatalog, to ensure a more comprehensive SV catalog that can inform GAL in diverse populations.

The CFIT sequencing data used in this study is described in Eckford et al. (2019) and is available through the Canadian CF registry at https://www.cysticfibrosis.ca/our-programs/cf-registry/requesting-canadian-cf-registry-data. The datasets generated and/or analyzed during the current study are available in the Zenodo repository https://doi.org/10.5281/zenodo.13367574. GWAS SVatalog is available at https://svatalog.research.sickkids.ca/. The source code can be found at https://github.com/strug-hub/gwas-svatalog. The documentation for the tool is available at https://gwas-svatalog-docs.readthedocs.io/en/latest/index.html.

• 10XGenomics (2020) longranger: 10x Genomics Linked-Read Alignment, Variant Calling, Phasing, and Structural Variant Calling https://github.com/10XGenomics/longranger

• 1000 Genomes Project Consortium, Auton A, Brooks LD, Durbin RM, Garrison EP, Kang HM et al. (2015) A global reference for human genetic variation. Nature 526:68–74.

  Article  Google Scholar 

• Abyzov A, Urban AE, Snyder M, Gerstein M (2011) CNVnator: an approach to discover, genotype, and characterize typical and atypical CNVs from family and population genome sequencing. Genome Res 21:974–984.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• All of Us Research Program Genomics Investigators (2024) Genomic data in the All of Us Research Program. Nature 627:340–346.

  Article  Google Scholar 

• Audano PA, Sulovari A, Graves-Lindsay TA, Cantsilieris S, Sorensen M, Welch AE et al. (2019) Characterizing the major structural variant alleles of the human genome. Cell 176:663–675.e19.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Benyamin B, McRae AF, Zhu G, Gordon S, Henders AK, Palotie A et al. (2009) Variants in TF and HFE explain approximately 40% of genetic variation in serum-transferrin levels. Am J Hum Genet 84:60–65.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Beyter D, Ingimundardottir H, Oddsson A, Eggertsson HP, Bjornsson E, Jonsson H et al. (2021) Long-read sequencing of 3,622 Icelanders provides insight into the role of structural variants in human diseases and other traits. Nat Genet 53:779–786.

  Article  CAS  PubMed  Google Scholar 

• Blanco-Rojo R, Bayele HK, Srai SKS, Vaquero MP (2012) Intronic SNP rs3811647 of the human transferrin gene modulates its expression in hepatoma cells. Nutr Hosp 27:2142–2145.

  CAS  PubMed  Google Scholar 

• Burrows JF, McGrattan MJ, Johnston JA (2005) The DUB/USP17 deubiquitinating enzymes, a multigene family within a tandemly repeated sequence. Genomics 85:524–529.

  Article  CAS  PubMed  Google Scholar 

• Chan AJS, Engchuan W, Reuter MS, Wang Z, Thiruvahindrapuram B, Trost B et al. (2022) Genome-wide rare variant score associates with morphological subtypes of autism spectrum disorder. Nat Commun 13: 6463.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Chen S, Francioli LC, Goodrich JK, Collins RL, Kanai M, Wang Q, et al. (2022). A genome-wide mutational constraint map quantified from variation in 76,156 human genomes. BioRxiv.

• Chen S, Francioli LC, Goodrich JK, Collins RL, Kanai M, Wang Q et al. (2024) A genomic mutational constraint map using variation in 76,156 human genomes. Nature 625:92–100.

  Article  CAS  PubMed  Google Scholar 

• Chen LJ, Li FF, Lu SY, Zhang XJ, Kam KW, Tang SM et al. (2021) Association of polymorphisms in ZFHX1B, KCNQ5 and GJD2 with myopia progression and polygenic risk prediction in children. Br J Ophthalmol 105:1751–1757.

  Article  PubMed  Google Scholar 

• Chen X, Schulz-Trieglaff O, Shaw R, Barnes B, Schlesinger F, Källberg M et al. (2016) Manta: rapid detection of structural variants and indels for germline and cancer sequencing applications. Bioinformatics 32:1220–1222.

  Article  CAS  PubMed  Google Scholar 

• Chiang C, Scott AJ, Davis JR, Tsang EK, Li X, Kim Y et al. (2017) The impact of structural variation on human gene expression. Nat Genet 49:692–699.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Collins RL, Brand H, Karczewski KJ, Zhao X, Alföldi J, Francioli LC et al. (2020) A structural variation reference for medical and population genetics. Nature 581:444–451.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Coutelier M, Holtgrewe M, Jäger M, Flöttman R, Mensah MA, Spielmann M et al. (2022) Combining callers improves the detection of copy number variants from whole-genome sequencing. Eur J Hum Genet 30:178–186.

  Article  CAS  PubMed  Google Scholar 

• Danecek P, Bonfield JK, Liddle J, Marshall J, Ohan V, Pollard MO, et al. (2021). Twelve years of SAMtools and BCFtools. Gigascience 10.

• Delaneau O, Zagury J-F, Robinson MR, Marchini JL, Dermitzakis ET (2019) Accurate, scalable and integrative haplotype estimation. Nat Commun 10: 5436.

  Article  PubMed  PubMed Central  Google Scholar 

• Eckford PDW, McCormack J, Munsie L, He G, Stanojevic S, Pereira SL et al. (2019) The CF Canada-Sick Kids Program in individual CF therapy: A resource for the advancement of personalized medicine in CF. J Cyst Fibros 18:35–43.

  Article  PubMed  Google Scholar 

• English AC, Menon VK, Gibbs RA, Metcalf GA, Sedlazeck FJ (2022) Truvari: refined structural variant comparison preserves allelic diversity. Genome Biol 23: 271.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Fritsche LG, Loenhardt T, Janssen A, Fisher SA, Rivera A, Keilhauer CN et al. (2008) Age-related macular degeneration is associated with an unstable ARMS2 (LOC387715) mRNA. Nat Genet 40:892–896.

  Article  CAS  PubMed  Google Scholar 

• Gaunt TR, Rodriguez S, Zapata C, Day INM (2006) MIDAS: software for analysis and visualisation of interallelic disequilibrium between multiallelic markers. BMC Bioinforma 7: 227.

  Article  Google Scholar 

• Broad Institute (2019): Picard Toolkit – A set of command line tools (in Java) for manipulating high-throughput sequencing (HTS) data and formats such as SAM/BAM/CRAM and VCF. https://github.com/broadinstitute/picard.

• Gong J, Wang F, Xiao B, Panjwani N, Lin F, Keenan K et al. (2019) Genetic association and transcriptome integration identify contributing genes and tissues at cystic fibrosis modifier loci. PLoS Genet 15:e1008007.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Gudmundsson S, Singer-Berk M, Watts NA, Phu W, Goodrich JK, Solomonson M et al. (2022) Variant interpretation using population databases: Lessons from gnomAD. Hum Mutat 43:1012–1030.

  Article  PubMed  Google Scholar 

• Günther T, Nettelblad C (2019) The presence and impact of reference bias on population genomic studies of prehistoric human populations. PLoS Genet 15:e1008302.

  Article  PubMed  PubMed Central  Google Scholar 

• Hancks DC, Kazazian HH (2010) SVA retrotransposons: Evolution and genetic instability. Semin Cancer Biol 20:234–245.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Hedrick PW (1987) Gametic disequilibrium measures: proceed with caution. Genetics 117:331–341.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Hong X, Roh W, Sullivan RJ, Wong KHK, Wittner BS, Guo H et al. (2021) The lipogenic regulator SREBP2 induces transferrin in circulating melanoma cells and suppresses ferroptosis. Cancer Discov 11:678–695.

  Article  CAS  PubMed  Google Scholar 

• Hu Y, Sun J-Y, Zhang Y, Zhang H, Gao S, Wang T et al. (2021) rs1990622 variant associates with Alzheimer’s disease and regulates TMEM106B expression in human brain tissues. BMC Med 19: 11.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Jakobsdottir J, Conley YP, Weeks DE, Mah TS, Ferrell RE, Gorin MB (2005) Susceptibility genes for age-related maculopathy on chromosome 10q26. Am J Hum Genet 77:389–407.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Jiang Y, Chen S, Wang X, Liu M, Iacono WG, Hewitt JK(2020) Association Analysis and Meta-Analysis of Multi-Allelic Variants for Large-Scale Sequence Data. Genes 11:586.

  Article  PubMed  PubMed Central  Google Scholar 

• Jin Y, Schaffer AA, Feolo M, Holmes JB, Kattman BL (2019) GRAF-pop: A Fast Distance-Based Method To Infer Subject Ancestry from Multiple Genotype Datasets Without Principal Components Analysis. G3 (Bethesda) 9:2447–2461.

  Article  PubMed  Google Scholar 

• Kosugi S, Terao C (2024) Comparative evaluation of SNVs, indels, and structural variations detected with short- and long-read sequencing data. Hum Gen Var 11:18.

  Article  CAS  Google Scholar 

• Landrum MJ, Chitipiralla S, Brown GR, Chen C, Gu B, Hart J et al. (2020) ClinVar: improvements to accessing data. Nucleic Acids Res 48:D835–D844.

  Article  CAS  PubMed  Google Scholar 

• Lappalainen I, Lopez J, Skipper L, Hefferon T, Spalding JD, Garner J et al. (2013) DbVar and DGVa: public archives for genomic structural variation. Nucleic Acids Res 41:D936–D941.

  Article  CAS  PubMed  Google Scholar 

• Lee JY, Harney DJ, Teo JD, Kwok JB, Sutherland GT, Larance M et al. (2023) The major TMEM106B dementia risk allele affects TMEM106B protein levels, fibril formation, and myelin lipid homeostasis in the ageing human hippocampus. Mol Neurodegener 18:63.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Li FF, Lu SY, Tang SM, Kam KW, Pancy OST, Yip WWK et al. (2021) Genetic associations of myopia severities and endophenotypes in children. Br J Ophthalmol 105:1178–1183.

  Article  PubMed  Google Scholar 

• Liao W-W, Asri M, Ebler J, Doerr D, Haukness M, Hickey G et al. (2023) A draft human pangenome reference. Nature 617:312–324.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Liao X, Yap MKH, Leung KH, Kao PYP, Liu LQ, Yip SP (2017) Genetic Association Study of KCNQ5 Polymorphisms with High Myopia. Biomed Res Int 2017:3024156.

  Article  PubMed  PubMed Central  Google Scholar 

• Logsdon GA, Ebert P, Audano PA, Loftus M, Porubsky D, Ebler J, et al. (2024). Complex genetic variation in nearly complete human genomes. BioRxiv.

• MacDonald JR, Ziman R, Yuen RKC, Feuk L, Scherer SW (2014) The Database of Genomic Variants: a curated collection of structural variation in the human genome. Nucleic Acids Res 42:D986–D992.

  Article  CAS  PubMed  Google Scholar 

• Mahmoud M, Gobet N, Cruz-Dávalos DI, Mounier N, Dessimoz C, Sedlazeck FJ (2019) Structural variant calling: the long and the short of it. Genome Biol 20: 246.

  Article  PubMed  PubMed Central  Google Scholar 

• Mahmoud M, Huang Y, Garimella K, Audano PA, Wan W, Prasad N et al. (2024) Utility of long-read sequencing for All of Us. Nat Commun 15: 837.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Martiniano R, Garrison E, Jones ER, Manica A, Durbin R (2020) Removing reference bias and improving indel calling in ancient DNA data analysis by mapping to a sequence variation graph. Genome Biol 21: 250.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Mastromatteo S, Chen A, Gong J, Lin F, Thiruvahindrapuram B, Sung WWL et al. (2023) High-quality read-based phasing of cystic fibrosis cohort informs genetic understanding of disease modification. HGG Adv 4:100156.

  CAS  PubMed  Google Scholar 

• Maurano MT, Humbert R, Rynes E, Thurman RE, Haugen E, Wang H et al. (2012) Systematic localization of common disease-associated variation in regulatory DNA. Science 337:1190–1195.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• McArthur E, Capra JA (2021) Topologically associating domain boundaries that are stable across diverse cell types are evolutionarily constrained and enriched for heritability. Am J Hum Genet 108:269–283.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Meynert AM, Ansari M, FitzPatrick DR, Taylor MS (2014) Variant detection sensitivity and biases in whole genome and exome sequencing. BMC Bioinforma 15:247.

  Article  Google Scholar 

• Micklisch S, Lin Y, Jacob S, Karlstetter M, Dannhausen K, Dasari P et al. (2017) Age-related macular degeneration associated polymorphism rs10490924 in ARMS2 results in deficiency of a complement activator. J Neuroinflammation 14:4.

  Article  PubMed  PubMed Central  Google Scholar 

• Morales J, Pujar S, Loveland JE, Astashyn A, Bennett R, Berry A et al. (2022) A joint NCBI and EMBL-EBI transcript set for clinical genomics and research. Nature 604:310–315.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Nassar LR, Barber GP, Benet-Pagès A, Casper J, Clawson H, Diekhans M et al. (2023) The UCSC Genome Browser database: 2023 update. Nucleic Acids Res 51:D1188–D1195.

  Article  CAS  PubMed  Google Scholar 

• Orozco G, Schoenfelder S, Walker N, Eyre S, Fraser P (2022) 3D genome organization links non-coding disease-associated variants to genes. Front Cell Dev Biol 10:995388.

  Article  PubMed  PubMed Central  Google Scholar 

• PacificBiosciences (2018): pbsv – PacBio structural variant (SV) calling and analysis tools https://github.com/PacificBiosciences/pbsv.

• PacificBiosciences (2020): pbmm2 – A minimap2 frontend for PacBio native data formats https://github.com/PacificBiosciences/pbmm2.

• Panigrahi A, O’Malley BW (2021) Mechanisms of enhancer action: the known and the unknown. Genome Biol 22: 108.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Panjwani N, Wang F, Mastromatteo S, Bao A, Wang C, He G et al. (2020) LocusFocus: Web-based colocalization for the annotation and functional follow-up of GWAS. PLoS Comput Biol 16:e1008336.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Pichler I, Minelli C, Sanna S, Tanaka T, Schwienbacher C, Naitza S et al. (2011) Identification of a common variant in the TFR2 gene implicated in the physiological regulation of serum iron levels. Hum Mol Genet 20:1232–1240.

  Article  CAS  PubMed  Google Scholar 

• Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MAR, Bender D et al. (2007) PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 81:559–575.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Quinn JP, Bubb VJ (2014) SVA retrotransposons as modulators of gene expression. Mob Genet Elem 4:e32102.

  Article  Google Scholar 

• Rehm HL, Page AJH, Smith L, Adams JB, Alterovitz G, Babb LJ, et al. (2021). GA4GH: International policies and standards for data sharing across genomic research and healthcare. Cell Genomics 1.

• Ren C, Zhang J, Yan W, Zhang Y, Chen X (2016) RNA-binding Protein PCBP2 Regulates p73 Expression and p73-dependent Antioxidant Defense. J Biol Chem 291:9629–9637.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Rivera A, Fisher SA, Fritsche LG, Keilhauer CN, Lichtner P, Meitinger T et al. (2005) Hypothetical LOC387715 is a second major susceptibility gene for age-related macular degeneration, contributing independently of complement factor H to disease risk. Hum Mol Genet 14:3227–3236.

  Article  CAS  PubMed  Google Scholar 

• Schloissnig S, Pani S, Ebler J, Hain C, Tsapalou V, Söylev A, et al. (2025). Structural variation in 1,019 diverse humans based on long-read sequencing. Nature.

• Sedlazeck FJ, Rescheneder P, Smolka M, Fang H, Nattestad M, von, Haeseler A et al. (2018) Accurate detection of complex structural variations using single-molecule sequencing. Nat Methods 15:461–468.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Sherry ST, Ward MH, Kholodov M, Baker J, Phan L, Smigielski EM et al. (2001) dbSNP: the NCBI database of genetic variation. Nucleic Acids Res 29:308–311.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Sollis E, Mosaku A, Abid A, Buniello A, Cerezo M, Gil L et al. (2023) The NHGRI-EBI GWAS Catalog: knowledgebase and deposition resource. Nucleic Acids Res 51:D977–D985.

  Article  CAS  PubMed  Google Scholar 

• Stachowska L, Koziarska D, Karakiewicz B, Kotwas A, Knyszyńska A, Folwarski M, et al. (2022). Hepcidin (rs10421768), Transferrin (rs3811647, rs1049296) and Transferrin Receptor 2 (rs7385804) Gene Polymorphism Might Be Associated with the Origin of Multiple Sclerosis. Int J Environ Res Public Health 19.

• Trost B, Thiruvahindrapuram B, Chan AJS, Engchuan W, Higginbotham EJ, Howe JL et al. (2022) Genomic architecture of autism from comprehensive whole-genome sequence annotation. Cell 185:4409–4427.e18.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Uffelmann E, Huang QQ, Munung NS, de Vries J, Okada Y, Martin AR et al. (2021) Genome-wide association studies. Nat Rev Methods Prim 1:59.

  Article  CAS  Google Scholar 

• Valiente-Mullor C, Beamud B, Ansari I, Francés-Cuesta C, García-González N, Mejía L et al. (2021) One is not enough: On the effects of reference genome for the mapping and subsequent analyses of short-reads. PLoS Comput Biol 17:e1008678.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Van Deerlin VM, Sleiman PMA, Martinez-Lage M, Chen-Plotkin A, Wang L-S, Graff-Radford NR et al. (2010) Common variants at 7p21 are associated with frontotemporal lobar degeneration with TDP-43 inclusions. Nat Genet 42:234–239.

  Article  PubMed  PubMed Central  Google Scholar 

• Van der Auwera G, O’Connor B (2020). Genomics in the Cloud: Using Docker, GATK, and WDL in Terra, 1st edn. O’Reilly Media: Sebastopol, CA.

• Verhoeven VJM, Hysi PG, Wojciechowski R, Fan Q, Guggenheim JA, Höhn R et al. (2013) Genome-wide meta-analyses of multiancestry cohorts identify multiple new susceptibility loci for refractive error and myopia. Nat Genet 45:314–318.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Vialle RA, de Paiva Lopes K, Bennett DA, Crary JF, Raj T (2022) Integrating whole-genome sequencing with multi-omic data reveals the impact of structural variants on gene regulation in the human brain. Nat Neurosci 25:504–514.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Visscher PM, Wray NR, Zhang Q, Sklar P, McCarthy MI, Brown MA et al. (2017) 10 years of GWAS discovery: biology, function, and translation. Am J Hum Genet 101:5–22.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Wang K, Li M, Hakonarson H (2010) ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res 38:e164.

  Article  PubMed  PubMed Central  Google Scholar 

• Yang G-F, Zhang X, Su Y-G, Zhao R, Wang Y-Y (2021) The role of the deubiquitinating enzyme DUB3/USP17 in cancer: a narrative review. Cancer Cell Int 21: 455.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Yu Z, Feng J, Wang W, Deng Z, Zhang Y, Xiao L et al. (2020) The EGFR-ZNF263 signaling axis silences SIX3 in glioblastoma epigenetically. Oncogene 39:3163–3178.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Zapata C (2000) The D’ measure of overall gametic disequilibrium between pairs of multiallelic loci. Evolution 54:1809–1812.

  CAS  PubMed  Google Scholar 

• Zarrei M, MacDonald JR, Merico D, Scherer SW (2015) A copy number variation map of the human genome. Nat Rev Genet 16:172–183.

  Article  CAS  PubMed  Google Scholar 

• Zhao H, Nettleton D, Dekkers JCM (2007) Evaluation of linkage disequilibrium measures between multi-allelic markers as predictors of linkage disequilibrium between single nucleotide polymorphisms. Genet Res 89:1–6.

  Article  CAS  PubMed  Google Scholar 

• Zhu M, Need AC, Han Y, Ge D, Maia JM, Zhu Q et al. (2012) Using ERDS to infer copy-number variants in high-coverage genomes. Am J Hum Genet 91:408–421.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

• Zook JM, Catoe D, McDaniel J, Vang L, Spies N, Sidow A et al. (2016) Extensive sequencing of seven human genomes to characterize benchmark reference materials. Sci Data 3: 160025.

  Article  CAS  PubMed  PubMed Central  Google Scholar 

Download references

Authors and Affiliations

1. Program in Genetics and Genome Biology, The Hospital for Sick Children, Toronto, ON, Canada

   Shalvi Chirmade, Zhuozhi Wang, Scott Mastromatteo, Eric Sanders, Bhooma Thiruvahindrapuram, Thomas Nalpathamkalam, Giovanna Pellecchia, Fan Lin, Katherine Keenan, Rohan V. Patel, Delnaz Roshandel, Joe Whitney, Sana Allana & Lisa J. Strug

2. The Centre for Applied Genomics, The Hospital for Sick Children, Toronto, ON, Canada

   Shalvi Chirmade, Zhuozhi Wang, Scott Mastromatteo, Bhooma Thiruvahindrapuram, Thomas Nalpathamkalam, Giovanna Pellecchia, Rohan V. Patel, Delnaz Roshandel, Joe Whitney, Sana Allana & Lisa J. Strug

3. Biostatistics Division, Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada

   Eric Sanders & Lisa J. Strug

4. Department of Pediatric Laboratory Medicine, The Hospital for Sick Children, Toronto, ON, Canada

   Wilson WL Sung

5. Translational Medicine, The Hospital for Sick Children, Toronto, ON, Canada

   Julie Avolio & Felix Ratjen

6. Cystic Fibrosis Canada, Toronto, ON, Canada

   Paul DW Eckford

7. Departments of Statistical Sciences and Computer Science, University of Toronto, Toronto, ON, Canada

   Lisa J. Strug

8. Department of Paediatrics, The Hospital for Sick Children, Toronto, ON, Canada

   Felix Ratjen

Contributions

Conceptualization: L.J.S. Sample recruitment: K.K., J.A., P.E., F.R. Sample processing: F.L., K.K. Data processing: S.C., Z.W., S.M., B.T., T.M., G.P., R.P., W.S., S.A., J.W. Formal analysis: S.C., Z.W. Funding acquisition: L.J.S. Investigation: S.C., Z.W., B.T., S.M., E.S., L.J.S. Methodology: S.C., Z.W. Project administration: L.J.S. Supervision: L.J.S. Visualization: S.C. Writing-original draft: S.C., Z.W., L.J.S. Writing, review, and editing: all authors.

Corresponding author

Correspondence to Lisa J. Strug.

Competing interests

The authors declare no competing interests.

Ethics

SickKids Program in Individualized CF Therapy (CFIT) was approved by the Research Ethics Board of the Hospital for Sick Children (#1000044783 from 2014 to present) and sub-site St. Michael’s Hospital (#14 188 from 2015 to present). Written informed consent was obtained from all participants or parents/guardians/substitute decision makers prior to inclusion in the study.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Associate editor: Megan Dennis.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions