Medicare's 2018 NCD on cancer testing governs uses of "next generation" sequencing. With their proposal to create a new modifier with the same terminology (here), do we need to define NGS? Or is it everything post Sanger, post 2005?
I asked Chat GPT.
##
This is a very perceptive question — and one that people in genomics quietly roll their eyes about at conferences.
The Short Answer
“Next-generation sequencing” (NGS) has become a frozen historical term. It no longer literally means “one generation after Sanger.”
It now functions more like a brand label than a generational description.
Chain termination chemistry
-
One fragment at a time
-
Capillary electrophoresis
-
High accuracy, low throughput
-
Dominated clinical sequencing through ~2005
When people said “first generation,” they meant Sanger
Second Generation = Next Generation
Platforms like Illumina changed the model:
-
Massive parallelization
-
Millions–billions of short reads
-
Computational reassembly
-
Cost collapse
“Next-generation” originally meant “post-Sanger short-read massively parallel sequencing.”
But that was 2006.
And yet… we still call it NGS.
What Happened to “Third-Generation” Sequencing?
For a while (~2012–2018), people used “third-generation sequencing” to describe:
Single-molecule sequencing
-
Long-read platforms
-
Real-time detection
-
No amplification
Companies like:
-
Pacific Biosciences (SMRT sequencing)
-
Oxford Nanopore Technologies
These were widely called “third-generation” for several years.
But the term never stuck commercially or clinically.
Why?
-
Marketing inertia – “NGS” became the regulatory, reimbursement, and CPT language.
-
Policy codification – CMS NCD 90.2 literally uses “NGS.”
-
Platform blending – Long-read, short-read, hybrid assemblies — the boundaries blurred.
-
No dramatic conceptual rupture like the shift from Sanger to massively parallel sequencing.
So “third-generation” faded.
What Do We Call Things Now?
Today, people use:
-
NGS (generic umbrella term)
-
Short-read sequencing
-
Long-read sequencing
-
Single-molecule sequencing
-
Spatial genomics
-
Transcriptomics
-
Multiomics
The field shifted from “generation labels” to read-length and application descriptors.
Will There Be a “Fourth Generation”?
Probably — but it won’t be called that.
The next real conceptual rupture would involve something like:
-
Direct molecular sensing without base calling
-
Native proteogenomic sequencing
-
In situ sequencing at cellular resolution
-
AI-first signal interpretation replacing traditional base assembly
If that happens, we may not call it “4th generation sequencing.”
We might call it:
-
Molecular sensing
-
Real-time omics
-
Embedded biosensing
-
Continuous genomic monitoring
Or something marketing-friendly and unhelpful.
Why NGS Stayed Frozen
This is similar to how:
-
“Personal computers” are still PCs
-
“Digital cameras” stayed digital
-
“Artificial intelligence” now means everything
“Next-generation sequencing” is no longer generational — it’s institutional.
Regulators, CPT panels, investors, journals — they all embedded the term.
You, especially in policy work around NCD 90.2, see this directly: CMS doesn’t want to adjudicate “what generation” a test is. They adjudicate whether it qualifies as NGS under their definition.
The Reality in 2026
-
NGS = any high-throughput DNA sequencing platform
-
The generational framing is obsolete.
-
The field evolved past the terminology.
“Third-generation sequencing” didn’t vanish because it was wrong — it vanished because the ecosystem stopped needing generational labels.
