A genome-wide deletion map in 125,730 individuals for novel rare disease gene and variant discovery

By analyzing homozygous deletions across 125,730 individuals in the UK's National Genomic Research Library, this study establishes a comprehensive genome-wide map that enhances rare disease diagnosis, uncovers the significance of non-coding promoter deletions, and identifies 43 candidate novel disease-associated genes.

The Gist

Imagine your DNA as a massive, 3-billion-page instruction manual for building and running a human being. Most of the time, we read these instructions perfectly. But sometimes, pages go missing. In the world of genetics, these missing pages are called deletions.

This paper is like a massive "missing page" detective story. The researchers looked at the genetic instruction manuals of 125,730 people (a huge crowd!) to find exactly where pages were missing. They weren't just looking for small typos; they were hunting for whole paragraphs or even entire chapters that had been ripped out.

Here is the story of what they found, broken down into simple parts:

1. The "Natural Knockout" Map

The researchers built a giant map showing every single place in the human genome where a page could be completely missing without the person getting sick.

• The Analogy: Think of it like testing a car engine by removing different parts. If you remove the radio, the car still drives. If you remove the engine, it stops.
• The Discovery: They found that humans are surprisingly tough. About 7% of the entire instruction manual can be missing, and the person is still healthy. This means our bodies have a lot of "spare parts" or redundant instructions. However, they also found that if you rip out pages from the "engine room" (protein-coding genes), people often get rare diseases.

2. Solving the Unsolved Mysteries

About half of the people with rare diseases in this study still didn't have a diagnosis, even after standard genetic testing. Why? Because standard tests are like looking for a missing word in a sentence; they often miss if the whole sentence is gone.

• The Fix: The researchers used a special trick. They looked for "silence" in the data. If a specific section of DNA usually has a lot of noise (reads) but suddenly goes completely silent in a patient, it means that section is missing.
• The Result: They found 295 people who finally got a diagnosis. The missing pages were in genes known to cause disease, but previous tests had missed them.
• The "Hidden" Clue: They also found that sometimes the missing page isn't the main instruction, but the cover page (the promoter or 5' UTR). If the cover is gone, the book never gets opened, even if the pages inside are fine. They found 19 cases where this "cover page deletion" was the cause of the disease.

3. Finding New "Bad Guys" (Novel Genes)

The most exciting part was finding genes that nobody knew were dangerous before.

• The Method: They looked for genes that were missing in at least two different people who had the exact same symptoms. It's like finding two different houses with the same broken window and realizing, "Hey, that window type is the problem!"
• The Three New Suspects:
  1. PDC (The Eye Guardian): They found four people with a missing PDC gene who all had severe, early blindness (Leber Congenital Amaurosis). It turns out this gene is crucial for how our eyes process light. The researchers suspect this missing piece might be a "family heirloom" mutation common in the UK, passed down from an ancestor who lived 600–1,000 years ago.
  2. GCG (The Stomach & Brain Link): They found two sisters with a missing GCG gene. They had intellectual disabilities and severe tummy trouble (diarrhea) as babies. This gene makes hormones that talk to both the brain and the gut. The researchers think this explains why these sisters had both brain and stomach issues.
  3. ENTPD3 (The Brain's Social Switch): They found three people with missing ENTPD3 genes who all had intellectual disabilities and autism. This gene is very active in the brain, suggesting it plays a key role in how our brains develop and socialize.

4. Why This Matters (According to the Paper)

The paper concludes that looking for these "missing pages" is a powerful way to solve medical mysteries that other tests miss.

• The "Silent" Success: They found that 0.5% of the rare disease patients in their study were solved only because they looked for these big deletions.
• The "Founder" Effect: They showed that some of these missing pages are common in specific areas (like North-West England), meaning doctors in those areas might see more of these specific conditions.
• The Future: By mapping out which parts of the manual can be safely removed (the 7% tolerance map), scientists can better understand which genes are safe to "turn off" for future therapies, and which ones are too dangerous to touch.

In short: The researchers created a giant map of missing genetic pages. This map helped them solve hundreds of medical mysteries, revealed that "cover page" deletions are a hidden cause of disease, and discovered three new genes that, when missing, cause blindness, gut-brain issues, and autism.

Technical Summary

Technical Summary: A Genome-Wide Deletion Map for Rare Disease Gene and Variant Discovery

Problem Statement
Structural variants (SVs), particularly homozygous deletions, are a significant but often under-detected cause of rare genetic disorders. While short-read genome sequencing is the standard for clinical genetic testing, it suffers from inherent limitations in SV detection, including high false-positive rates and missed variants, particularly in repetitive or GC-rich regions. Consequently, approximately half of rare disease patients remain without a genetic diagnosis. Existing methods often struggle to distinguish true homozygous deletions from artifacts, and the tolerance of the human genome to complete gene loss (homozygous knockout) remains poorly characterized on a large scale. Furthermore, non-coding deletions, such as those affecting promoters or 5'UTRs, are frequently overlooked in clinical pipelines despite their potential to abolish gene transcription.

Methodology
The authors leveraged genome sequencing data from 125,730 individuals within the UK's National Genomic Research Library (NGRL), comprising participants from the Genomics England 100,000 Genomes Project and the NHS Genomic Medicine Service. The cohort included 58,022 individuals with rare disorders, 50,484 unaffected family members, and 17,224 individuals recruited for cancer.

To address the noise inherent in SV calling, the authors developed a computational approach focusing on the distinct read-depth signal associated with homozygous deletions. The pipeline involved:

1. Data Aggregation: Integrating putative biallelic deletions identified by three different callers (Canvas, DRAGEN, and Manta).
2. High-Confidence Filtering: Defining high-confidence deletions based on stringent read-depth criteria: a median read-depth of ≤1 in the participant and ≥10 across a control sample (unaffected family members or germline cancer samples). This filtering significantly reduced noise, reducing the median number of deletions per individual from 1,781 to 464.
3. Variant Curation: Aggregating variants with >80% overlap and filtering for rare variants (cohort allele frequency ≤ 0.001).
4. Diagnostic and Discovery Analysis:
   • Diagnostic: Intersecting deletions with known disease genes (PanelApp 'green' and OMIM recessive lists) and assessing phenotypic fit. This included manual review in IGV and RNA-sequencing validation where available.
   • Novel Gene Discovery: Identifying genes without known disease associations that harbored biallelic deletions in at least two individuals with consistent phenotypes. This search space was restricted to genes with no homozygous deletions in control populations.
   • Orthogonal Validation: Expanding the search to include homozygous and compound heterozygous predicted loss-of-function (pLoF) SNVs/Indels in candidate genes to strengthen gene-disease associations.

Key Results

• Genome-Wide Tolerance Map: The study curated 535,699 high-confidence homozygous deletion SVs, of which 48,735 were rare. These deletions collectively covered 213 Mb (6.92% of the human genome). Notably, 3.11% of protein-coding bases were deleted in at least one individual, indicating substantial tolerance to complete sequence loss in specific genomic regions.
• Diagnostic Yield: In the subset of 58,022 rare disease individuals, the authors identified 295 individuals with likely diagnostic homozygous deletions impacting protein-coding regions of known disease genes. This represents a diagnostic rate of 0.5% for the cohort.
  • Non-Coding Promoter Deletions: A significant finding was the identification of 32 candidate non-coding SVs in or near known disease genes, 19 of which (59.37%) were recurrent or consistent with pathogenicity. These deletions disrupted 5'UTR or promoter regions, effectively abolishing transcription. RNA-sequencing validation confirmed dramatic transcript reduction in cases such as MICU1 and PLEC, highlighting promoter deletions as an underappreciated cause of rare disorders.
  • Consanguinity: Probands with diagnostic biallelic deletions were nearly nine-fold more likely to have consanguineous parents compared to the general cohort, consistent with a recessive inheritance model.
• Novel Gene Discovery: The authors identified 43 genes with no prior rare-disease association that harbored biallelic deletions in two or more individuals with consistent phenotypes. Three genes were highlighted in detail:
  • PDC (Phosducin): Four probands with homozygous 15kb deletions encompassing the entire gene presented with Leber Congenital Amaurosis. The deletion was identified as a potential UK founder variant.
  • GCG (Glucagon): Two sisters with a homozygous deletion of the entire GCG gene presented with intellectual disability and recurrent infantile diarrhea. A third individual with a homozygous stop-gain SNV in GCG showed a similar phenotype.
  • ENTPD3: Three probands with distinct homozygous deletions in ENTPD3 shared a phenotype of intellectual disability, microcephaly, and autism.

Significance and Claims
The paper claims to provide the first genome-wide map of homozygous deletions derived from short-read sequencing data at this scale, demonstrating the utility of read-depth signals for high-confidence SV detection. The study underscores that homozygous deletions contribute to approximately 0.5% of rare disease diagnoses in the NGRL cohort, a figure that may be an underestimate due to prior negative testing with microarrays.

Crucially, the authors assert that their approach reveals "promoter deletion as an underappreciated cause of rare disorders," demonstrating that non-coding deletions can be as pathogenic as coding truncations. By identifying novel candidate disease genes (PDC, GCG, ENTPD3) and characterizing genome-wide tolerance to loss, the dataset serves as a resource for both improving rare disease diagnosis and informing therapeutic strategies (e.g., identifying genes tolerant to knockdown). The authors conclude that while limitations regarding false positives and the inability to detect very large SVs with short reads exist, this high-confidence dataset significantly advances the understanding of the genomic burden of homozygous deletions in human disease.