Map of 1.6 Million Structural Variants From 467,000 Genomes Yields 28,000 Health Associations

Genome research has long focused on variants that change a single base. Structural variants (SVs), in which chunks of more than 50 base pairs are deleted, duplicated or inverted, can have large effects but have been studied less because they are hard to find. A team led by Slavé Petrovski at the AstraZeneca Centre for Genomics Research in the United Kingdom published a study in Cell that fills this gap at large scale.

A PheWAS that sweeps across phenotypes

The researchers identified 1.6 million structural variants in the whole genomes of 467,152 UK Biobank participants. They then ran a phenome-wide association study (PheWAS) examining associations with 13,322 disease presence/absence traits, 2,018 quantitative traits and 2,923 plasma protein levels. Of the structural variants found, 82.25% were outside the range that earlier copy-number variant studies could detect.

At the protein level, 3,467 structural variants were linked to blood levels of 1,690 proteins, and effects differed by the type and location of the variant. A total of 28,184 associations with health outcomes were detected, and some structural variants were rare but had large phenotypic effects. In addition, at locations known from genome-wide association studies (GWAS), 32.21% of the associated structural variants overlapped functional genomic regions, suggesting that structural variants may be the true cause of some GWAS signals.

Significance and limits

This map shows that the contribution of structural variants to human traits and disease has been underestimated, and it could be a resource for finding drug targets. However, UK Biobank participants are mostly middle-aged and older people of European ancestry, so the results are hard to apply directly to other populations, and association analysis does not prove causation. It is also worth noting that many of the authors work for a pharmaceutical company.