UK Biobank & Pharma Consortium + Illumina DRAGEN: 500,000 Whole-Genome Sequencing Variant Atlas
📌Key Takeaways
- 1UK Biobank Pharma Consortium (Amgen, AZ, GSK, J&J) (Population Genomics & Target Discovery, 490,640 Whole Genomes (32.5x coverage)) deployed Illumina DRAGEN.
- 2Per-Genome Secondary Analysis Turnaround: 50x faster per genome (now ~25 minutes (Illumina DRAGEN)).
- 3Total Human Genetic Variants Cataloged: 100x variant discovery (now 1.5+ Billion SNVs, indels, and SVs).
- 4Implementation timeline: 36 months sequencing & harmonized DRAGEN variant calling across 490,640 participants.
Scaling Beyond Exomes to Non-Coding & Structural Genome Space
Whole-exome sequencing captures only the 1.5% of the human genome that codes for proteins, missing 98.5% of regulatory enhancers, promoters, intronic splice modifiers, and large structural variants that drive complex cardiometabolic, autoimmune, and neurodegenerative traits. To build the world's largest open biomedical resource, a consortium of pharmaceutical companies (Amgen/deCODE, AstraZeneca, GSK, and Johnson & Johnson) funded the 32.5x whole-genome sequencing of all 500,000 UK Biobank participants. What's the takeaway for your own architecture? We've analyzed this rollout closely, and it's clear you can't separate compute from wet-lab execution. You'll get the best ROI when your scientists and modelers share one data loop.
High-Speed Harmonized Calling with Illumina DRAGEN & Graph Genomes
Processing half a million 32.5x whole genomes via legacy CPU BWA-GATK pipelines would have required over 12 million CPU-hours and produced inconsistent calls across repetitive regions. The consortium standardized on the Illumina DRAGEN pipeline, calling over 1.5 billion SNVs, indels, and structural variants with multigenome graph reference improvements. Hosted inside the DNAnexus-powered UK Biobank Research Analysis Platform (RAP), these harmonized VCF/pVCF calls link directly to 15 years of ICD-10 hospital records, Olink proteomics (3,000+ plasma proteins), and MRI imaging for genetically validated drug target discovery. What's the takeaway for your own architecture? We've analyzed this rollout closely, and it's clear you can't separate compute from wet-lab execution. You'll get the best ROI when your scientists and modelers share one data loop.
Per-Genome Secondary Analysis Turnaround
50x faster per genome
Total Human Genetic Variants Cataloged
100x variant discovery
PrecisionFDA Benchmark SNP & Indel F1 Score
80%+ error reduction
The Challenge
Analyzing 500,000 whole human genomes required petabyte-scale throughput, sub-30-minute per-sample turnaround, and high sensitivity in segmental duplications and structural variants. That's a massive bottleneck for any R&D team. If you're running similar programs, you know legacy workflows simply don't scale.
The Solution
Deployed Illumina DRAGEN secondary analysis and joint calling across deCODE Genetics and Wellcome Sanger Institute, served via the DNAnexus UK Biobank Research Analysis Platform. Here's how they fixed it. By pairing Illumina DRAGEN with automated wet-lab feedback, we see how you can compress months of manual screening into weeks.
Implementation
Timeline
36 months sequencing & harmonized DRAGEN variant calling across 490,640 participants
Team
deCODE Genetics, Wellcome Sanger Institute, Illumina, DNAnexus, and Pharma Consortium genetics teams
- 1Sequenced 490,640 participant blood samples on Illumina NovaSeq 6000 instruments to a mean depth of 32.5x.
- 2Executed harmonized alignment to GRCh38 and variant calling using Illumina DRAGEN to generate single-sample gVCFs and population pVCFs.
- 3Integrated structural variant (SV) and short tandem repeat (ExpansionHunter) callers alongside SNV and indel pipelines.
- 4Published the multi-petabyte dataset inside the DNAnexus cloud workspace linked to Olink plasma proteomics and longitudinal NHS health records.
Results
| Metric | Before | After | Change |
|---|---|---|---|
| Per-Genome Secondary Analysis Turnaround | 24+ hours (CPU BWA-MEM + GATK) | ~25 minutes (Illumina DRAGEN) | 50x faster per genome |
| Total Human Genetic Variants Cataloged | ~15M variants in exome-only releases | 1.5+ Billion SNVs, indels, and SVs | 100x variant discovery |
| PrecisionFDA Benchmark SNP & Indel F1 Score | 99.1% standard pipeline | 99.85%+ with DRAGEN Graph Genome | 80%+ error reduction |
"The release of whole-genome sequencing for all 500,000 UK Biobank participants represents a landmark moment for human genetics and genetically anchored drug discovery."
Prof. Sir Rory Collins, Principal Investigator & Chief Executive, UK Biobank
Key Learnings
- 1Hardware-accelerated secondary analysis (DRAGEN) is essential when scaling past 10,000 whole genomes to prevent compute bottlenecks.
- 2Graph-based reference genomes dramatically reduce false-positive variant calls in polymorphic HLA and pharmacogenomic loci.
- 3Bringing compute to the data inside a federated cloud workspace (DNAnexus RAP) eliminates multi-petabyte egress fees for academic and pharma researchers.
Frequently Asked Questions
Verified References & Citations (2)
- [1]
Whole-genome sequencing of 490,640 UK Biobank participants processed with Illumina DRAGEN reveals 1.5 billion variants.
- [2]
UK Biobank releases world's largest whole-genome sequencing dataset on DNAnexus Research Analysis Platform.