Key takeaways
- Sanger: one target per reaction, ~700–1,000 bases, very accurate. Best for a few known targets: clone checks, a point mutation, confirming a variant.
- NGS: millions of reads in parallel. Best for many targets or samples, whole genomes or transcriptomes, and low-frequency variants.
- Sanger misses variants below ~15–20% allele frequency. Deep NGS can see down to ~1%.
- Need the full sequence of a plasmid? Whole-plasmid long-read sequencing reads the entire plasmid from a single sample, with no primers. It is often a better fit than many Sanger reactions.
How the two methods differ
Sanger sequencing (capillary sequencing) reads one DNA region per reaction, starting from a primer you choose. The result is a chromatogram, a trace with one peak per base, and a single sequence of up to about 1,000 bases.
Next-generation sequencing (NGS) reads millions of DNA fragments in parallel and then puts them back together by alignment or assembly. Every position is read many times, which is what makes NGS sensitive and scalable.
| Sanger | NGS short-read (Illumina, Ultima) | NGS long-read (PacBio HiFi, Nanopore) | |
|---|---|---|---|
| Output | One sequence per reaction | Millions to billions of reads per run | Thousands to millions of long reads per run |
| Read length | ~700–1,000 bases | Typically 100–300 bases | Thousands to over 100,000 bases |
| Read quality | Q40+ (≥99.99%) in the middle of a good read; lower at the start and end | Most bases Q30+ (≥99.9%); quality drops toward the end of the read | PacBio HiFi Q30+ per read; Nanopore about Q20 (~99%) per raw read, higher after consensus |
| Typical errors | Mixed or noisy traces rather than random errors | Mostly single-base substitutions (Ultima: mostly homopolymer length) | Mostly small insertions or deletions, especially in homopolymers |
| Low-frequency variants | Detects down to ~15–20% | Detects down to ~1% with enough depth | Possible, usually at lower depth than short reads |
| Mixed templates | Overlapping, unreadable traces | Each molecule read separately | Each molecule read separately, end to end |
| Strengths | Quick checks of known targets | Many targets or samples, whole genomes, RNA-Seq | Whole plasmids, long amplicons, structural variants, repeats, de novo assembly |
| Data analysis | Simple: view the trace, align to a reference | Needs a bioinformatics pipeline | Needs a bioinformatics pipeline |
Read quality: what the numbers mean
Sequencing quality is usually given as a Phred score, written Q. Q20 means 1 error in 100 bases (99% accuracy), Q30 means 1 in 1,000 (99.9%) and Q40 means 1 in 10,000 (99.99%).
- Sanger gives Q40 or better through the middle of a clean read. The first few dozen bases after the primer and the far end of the read are less reliable, which is why important regions are read from both directions.
- Illumina reports most bases at Q30 or higher. Errors are mostly single-base substitutions and become more frequent toward the end of each read. Ultima Genomics reaches similar overall quality, but like other flow-based chemistries its errors are mostly in the length of homopolymers (runs of the same base).
- PacBio HiFi reads the same molecule many times and produces individual reads at Q30 or higher.
- Oxford Nanopore raw reads are around Q20 with current chemistry. Errors are mostly small insertions and deletions in homopolymers, and most are removed when many reads are combined into a consensus.
Two kinds of error behave differently. Random errors disappear as coverage goes up, because each position is read many times. Systematic errors, such as homopolymer length, can survive high coverage, so a critical base in such a region is worth confirming with a second method.
When Sanger is the right choice
- Clone and construct verification of an insert, junction or site-directed mutation.
- Confirming a variant that NGS has already found.
- Genotyping a few known positions in a small number of samples.
- Checking a PCR product before you move on.
In all of these you know what you expect to see, and a single clean read answers the question. You can send plasmids or PCR products in tubes or 96-well plates. GC-rich, repetitive or otherwise difficult templates may need a special reaction.
When NGS is the right choice
- Discovery: you don't know where the variants are (whole genome, whole exome).
- Many targets or many samples: tens to hundreds of loci across many samples (targeted amplicon sequencing).
- Low-frequency variants: tumour heterogeneity, mosaicism, edited cell pools, viral quasispecies.
- Mixed populations: CRISPR editing outcomes, microbiomes (16S and metagenomics).
- Expression: gene expression and transcript discovery (RNA-Seq).
A quick decision guide
| Your project | Usually the best fit |
|---|---|
| Check 5 clones for the right insert | Sanger |
| Verify the full sequence of a 7 kb plasmid | Whole-plasmid long-read sequencing |
| Confirm one variant found by exome sequencing | Sanger |
| Screen 200 samples at 30 loci | NGS targeted amplicon sequencing |
| Measure CRISPR editing efficiency in a cell pool | NGS amplicon sequencing |
| Find the causal variant in a patient or a mutant line | Whole exome or whole genome sequencing |
| Profile a gut or soil microbial community | 16S amplicon or shotgun metagenomics |
What about cost?
Sanger is priced per reaction, so it is the cheapest option for a handful of reads. NGS has a higher entry cost per run, but that cost is shared across every target and sample in the run. Once you multiply targets by samples into the hundreds, NGS almost always costs less per result. For a single plasmid, whole-plasmid long-read sequencing often costs less than the several Sanger reactions needed to walk across it.
QuickSeq shows an instant online estimate for Sanger and whole-plasmid sequencing. For NGS services, send us your project details and our team will reply with an estimate as soon as possible, so you can compare both routes before you decide.
Frequently asked questions
Is Sanger sequencing more accurate than NGS?
For a single clean template, Sanger is extremely accurate (around 99.99% per base) and is often used to confirm variants. NGS reaches similar or better consensus accuracy by reading each position many times, and it can also detect variants that Sanger misses because they are present in only a small fraction of the DNA.
What is the lowest variant frequency Sanger can detect?
Roughly 15–20% of the DNA in the sample. Below that, the second peak in the chromatogram is hard to distinguish from noise. Deep NGS can reliably detect variants down to about 1%, and lower with specialised methods.
How long a read do I get from Sanger sequencing?
Typically 700–1,000 high-quality bases per reaction, depending on the template. Longer regions need several reactions with different primers (primer walking).
When does NGS become cheaper than Sanger?
When you multiply targets by samples. A few reactions are cheapest with Sanger; tens to hundreds of amplicons, or many samples at once, usually cost less per target with NGS amplicon sequencing.
Can I verify a whole plasmid without Sanger primer walking?
Yes. Whole-plasmid long-read sequencing reads the full circular plasmid in one go, with no primers needed. See our plasmid verification guide.
