Despite three decades of NGS development, Sanger sequencing hasn’t gone away — it’s still the gold standard for confirming a single DNA sequence. The trick is knowing which experiment needs which approach.
How they work, briefly
Sanger sequencing uses chain-terminating dideoxynucleotides (ddNTPs) labeled with four fluorescent dyes. DNA polymerase extends a primer until it incorporates a ddNTP and stops. Capillary electrophoresis separates fragments by size, and a laser reads the dye at each position to produce an electropherogram.
NGS performs massively parallel sequencing of millions to billions of fragments simultaneously. Most platforms (Illumina) use sequencing-by-synthesis with reversible terminators. The result is millions of short reads (50–300 bp) that are aligned and assembled bioinformatically.
Side-by-side comparison
| Property | Sanger | NGS |
|---|---|---|
| Read length | 700–900 bp | 50–300 bp (Illumina); up to 100 kb+ (long read) |
| Throughput | 1 read per capillary | Millions to billions per run |
| Cost per base | High | Very low at scale |
| Cost per sample (single target) | Low | Higher |
| Detection of low-frequency variants | Limited (~15–20%) | Excellent (~1% or below) |
| Turnaround | Hours | Days |
| Bioinformatics required | Minimal | Substantial |
When Sanger is the right choice
- Confirming a plasmid construct after cloning
- Validating an NGS variant in a single patient or sample
- Genotyping a known mutation in a small number of samples
- Sequencing PCR products for species identification
- Quality control of CRISPR-edited clones at the target locus
When NGS is the right choice
- Discovery work — finding novel variants without prior knowledge
- Genome-scale studies — WGS, WES, transcriptomics, microbiome
- Detecting rare variants in heterogeneous samples (tumors, cell-free DNA)
- Sequencing many samples in parallel via multiplexing
- Single-cell or spatial work
The hybrid reality
Most labs use both. NGS finds candidates; Sanger confirms them. NGS profiles a population; Sanger reads the founding clone. NGS screens a CRISPR library; Sanger validates individual edits. Treating them as complementary instead of competitive is the right mental model.
Things that often surprise users
- Sanger has limited sensitivity — minor variants below ~15% are usually invisible
- NGS isn’t free — when you need to sequence one sample with one target, Sanger is faster and cheaper
- Long-read NGS (PacBio HiFi, Oxford Nanopore) blurs the boundary — you can now get accurate, very long reads
- Bioinformatics dominates NGS budgets — sequencing is the easy part
Sanger isn’t outdated — it’s specialized. NGS isn’t always better — it’s broader. Pick based on the question, the sample count, and the variant frequency you need to detect.
