Next-generation sequencing (NGS)
Also known as: NGS, high-throughput sequencing, massively parallel sequencing
Next-generation sequencing (NGS) refers to technologies that read the order of DNA letters in millions of fragments at the same time, rather than one fragment at a time.
It made reading a whole human genome go from a multi-billion-dollar, decade-long project to something done in about a day for a few hundred dollars. While NGS has transformed cancer diagnosis and rare disease testing, the hard part has shifted from reading DNA to interpreting what it means.
Next-generation sequencing (NGS), explained at three levels
At home
Imagine copying a huge book by shredding many copies into short strips, reading millions of strips at once, and then matching overlapping strips to rebuild the text. That is how next-generation sequencing reads DNA. The machine reads short fragments in parallel, and software lines them up against a reference genome to reconstruct a person's DNA and spot where it differs.
At work
Next-generation sequencing is a group of technologies that read millions of DNA fragments at once, rather than one at a time. It lets labs run broad genetic tests in a single run, from cancer panels to whole genomes. For hospitals and companies, it means faster diagnoses, targeted cancer treatment, outbreak tracking, and research at a small fraction of past costs.
At home
Labs choose how much DNA to read. A panel looks at a set of genes tied to one condition, such as hereditary breast cancer. An exome reads all protein-coding genes, about 1 to 2% of the genome but where most known disease-causing changes sit. A whole genome reads nearly everything, including the regions between genes. Wider tests find more, but they also turn up more uncertain results that are harder to interpret.
At work
A clinical NGS workflow runs in stages. Library preparation fragments DNA and adds short adapter sequences. Short-read instruments, dominated by Illumina, read fragments of about 150 letters through sequencing by synthesis, detecting fluorescent letters as they are added. Long-read platforms from PacBio and Oxford Nanopore read thousands of letters at once, handling repeats and structural changes. Bioinformatics pipelines then align reads to a reference, call variants, and annotate them before a geneticist classifies each one.
At home
Each genome holds millions of differences from the reference, and most are harmless. Labs classify variants on a five-tier scale, from benign to pathogenic, using population databases and published evidence. Many fall in the middle as variants of uncertain significance, and they can be reclassified years later. While more sequencing sounds better, in my view that uncertainty is one of the biggest sources of anxiety in genetic testing, and why genetic counselling matters.
At work
Scaling clinical NGS shifts cost from sequencing to everything around it. A whole genome can generate over 100 GB of raw data, so storage and compute matter. Labs need accreditation, validated pipelines, and ACMG-style variant classification. Reference databases under-represent many ancestries, increasing uncertain results for those patients. Policies must cover reanalysis, incidental findings, and consent, and Canada's Genetic Non-Discrimination Act limits who can require test results. Interpretation and counselling capacity are now the bottleneck.
Real-world examples
At home
Marcus's newborn niece had unexplained seizures. Rapid whole-genome sequencing at a children's hospital found a genetic cause within days, which changed her treatment immediately.
At work
Nadia is a lab director at a Canadian cancer centre. Her team uses NGS panels on tumour samples to find mutations that match targeted drugs, so oncologists can choose treatments based on each tumour's genetics.
How it compares
| feature | Next-generation sequencing (NGS) | Sanger sequencing |
|---|---|---|
| Throughput | Millions of fragments at once | One fragment at a time |
| Cost per genome | Hundreds of dollars | Impractical at genome scale |
| Best for | Genomes, panels, discovery | Confirming specific variants |
| Data volume | Very large | Small |
NGS reads genomes at scale. Sanger remains the gold standard for checking single spots.
For analysts
The cost of sequencing is no longer the bottleneck; interpretation and counselling capacity are. I'd budget for genetic counsellors and bioinformatics staff when expanding genomic testing, not just sequencing machines.
History & origin
Frederick Sanger developed his DNA sequencing method in 1977, and it powered the Human Genome Project, which finished a mostly complete human genome in 2003 at a cost of about US$2.7 billion. Massively parallel methods emerged in the mid-2000s from companies such as 454 Life Sciences and Solexa, which Illumina acquired in 2007. Costs fell faster than Moore's law, and Illumina announced a roughly US$1,000 genome in 2014. Long-read technologies from PacBio and Oxford Nanopore filled gaps, helping the Telomere-to-Telomere consortium publish the first truly complete human genome in 2022. Newer competitors such as Ultima Genomics and Element Biosciences pushed prices lower still. In my opinion, sequencing has become a commodity, and the value now lies in analysis, data sharing, and getting results to patients.