What Is Near-Perfect Genome Sequencing? The Next Evolution of Medical Genomics

Many rare genetic conditions are still investigated through a cascade of germline tests, from exome and short-read genome sequencing to specialized assays, often aimed at pinpointing a specific type of variation or designed for a particularly difficult-to-assess locus. Patients with rare diseases may undergo more than 10 tests before receiving an answer, and many remain undiagnosed.

At Radboud University Medical Center (Radboudumc), we have begun consolidating this fragmented pathway by using long-read genome sequencing as a frontline test. By reading much longer stretches of native DNA, long-reads can resolve complex variants, repetitive regions, and haplotypes that conventional approaches struggle to capture. The most advanced long-read methods, including HiFi sequencing, can also reveal epigenetic information linked to development and disease.

Our study shows what is possible for patients and laboratories when we stop using fragmented tests. But adopting long-read sequencing is only one part of the transformation for medical genetics. Near-perfect genome sequencing, or NPGS, combines long-read sequencing with diploid genome assembly, pangenome references, and AI-supported interpretation to improve diagnostics and deepen understanding of disease.

Moving to a one-test model

Long-read sequencing once offered premium insight at a premium cost. As costs have fallen and workflows matured, that insight has become accessible for routine clinical use. At Radboudumc, we set out to test whether one long-read genome could replace the existing cascade of assays without losing clinically relevant information. The prospectively designed pilot showed long-read sequencing could do more than simply match the established pathway—i.e., consolidate more than 10 standard-of-care tests. We actually saw long-read sequencing improve diagnostic performance by 3.4% across all included rare diseases. We modelled that, in our center alone, a complete switch to long-read genome sequencing as the only germline test would deliver a 2.5% gain, translating to around 400 more meaningful diagnoses for families each year.

Following the pilot, we introduced long-read sequencing as a genetic first-tier test in October 2025, initially for intellectual disability and neurodevelopmental diseases as well as visual impairment disorders. We have since expanded its use to non-coagulation and thrombotic disorders and, most recently, to hearing impairment in May 2026. A single test means fewer assays for our staff to run, analyze, and interpret. Crucially, it also means fewer rounds of testing for patients and families, and less time spent waiting for answers. Comprehensive genomic information also allows us to rule out a genetic diagnosis with greater confidence—sometimes, finding nothing is equally important.

To build on our progress and deliver the promise of NPGS, we will require further advances that change not only how we read the genome, but also how we reconstruct, compare, and interpret it. Here are four components of the NPGS framework:

1. Long-read genome sequencing: Capturing more of the biology in one test

Long-read genome sequencing is the foundation of NPGS. But laboratories do not need to invest in overhauling every diagnostic pathway at once. A practical starting point is in areas with broad genetic heterogeneity and frequent follow-on testing, such as neurodevelopmental disorders, multiple congenital anomalies, and unresolved rare disease cases. Usage can expand to other indications as sequencing costs continue to fall and throughput increases.

2. Diploid genome assembly: Reconstructing the genome a person actually inherited

Long-read sequencing is typically analyzed by aligning reads against a standard reference genome. But this method means that interpretation remains constrained by an external template that cannot fully represent every individual’s genomic structure. Diploid assembly independently reconstructs the two versions of the genome inherited from each parent. This produces continuous, phased sequences and shows which variants occur together on the same chromosome. In medical genetics, these insights provide a clearer view of complex rearrangements and improve the interpretation of conditions caused by combinations of variants.

3. Pangenome references: Comparing patients against human diversity

An accurate personal genome still needs to be compared with a reference genome that reflects human diversity. Today’s standard linear reference is Eurocentric and does not represent all populations equally.

Pangenomes replace a single linear reference with a framework containing many different human haplotypes. This broader comparison reduces reference bias and improves variant calling across ancestries. For example, in the Human Pangenome Reference Consortium’s draft, using a diverse reference reduced errors in identifying small genetic variants by 34% compared with the standard reference genome, with the greatest benefits for non-European ancestry. Therefore, using a pangenome in place of a standard reference genome should help laboratories distinguish genuinely rare disease-causing variants from benign variation that is simply absent from existing references.

4. AI-supported interpretation: Getting more answers from genomic data

As sequencing and assembly become comprehensive, the main challenge will increasingly move from variant identification to interpretation. NPGS will reveal variation in regions that have previously been difficult to analyze, creating a volume and complexity of findings that cannot be managed efficiently through manual interpretation alone.

AI can support several stages of this process, from extracting clinical features and prioritizing candidate variants to predicting functional effects, reviewing published evidence, and periodically reanalyzing unresolved cases. However, AI tools should support rather than replace clinical expertise, particularly while challenges remain regarding bias, factual errors, validation, and regulation.

The roadmap to NPGS

NPGS will not arrive through a single breakthrough or an overnight replacement of every assay. So, laboratories should not wait to begin consolidating tests. Start by targeting pathways where patients face repeated testing and interpretation creates the greatest burden.

While there is an upfront investment required in sequencing technology, workflows, and training, our experience at Radboudumc shows that this investment quickly pays off through fewer tests, efficient laboratory processes, and faster testing throughput. Yet the value goes far beyond cost: complete genomic data improves diagnosis and gives researchers a stronger foundation for understanding disease.

The remaining elements of NPGS can be added as the tools, infrastructure, and evidence mature. Achieving the full NPGS model will require collaboration across laboratories, clinicians, bioinformaticians, technology developers, and healthcare systems. But the first step is clear—stop treating each part of the genome as a separate diagnostic problem.