Sanger and NGS

Why it matters

Reading DNA sequence is now routine: confirming a plasmid clone, checking a patient variant, identifying a microbe, profiling a tumour or measuring expression of every gene at once. Different jobs need different technologies, and choosing badly wastes money or misses the answer. Technicians prepare samples for sequencing and often do the first quality checks on the data, so you need to know what each method can and cannot do.

Sanger sequencing

Sanger, or chain-termination, sequencing (1977) is still the standard for reading one short region with high accuracy.

1. A sequencing reaction resembles PCR with one primer, normal dNTPs and a small amount of fluorescently labelled dideoxynucleotidesLabelled nucleotides (ddNTPs) with no 3' hydroxyl, so a strand stops growing when one is added. (ddNTPs). A ddNTP has no 3' hydroxyl, so when the polymerase adds one, that strand stops growing.
2. The result is a mix of fragments ending at every position, each tagged with a colour for its last base.
3. Capillary electrophoresis separates fragments by length with single-base resolution, and a laser reads the colour of each as it passes.
4. The output is a chromatogramThe trace of coloured peaks produced by Sanger sequencing, with a base call for each peak. (trace) of coloured peaks, plus the base calls.

Strengths: very accurate, reads of about 700 to 1,000 bases, cheap and fast for a few samples. Limits: one template per reaction, so it does not scale to whole genomes, and it struggles to detect variants present in only a small fraction of cells (below roughly 15 to 20 percent).

Typical uses: confirming a cloned insert, checking a CRISPR edit at one site, confirming a variant found by NGS.

Next-generation sequencing (NGS)

NGS (also called massively parallel or high-throughput sequencing) reads millions to billions of DNA fragments at once. The most widespread approach, sequencing by synthesis from Illumina, works like this:

1. Library preparationFragmenting DNA and adding adapters, often with index barcodes, so it can be sequenced on an NGS run.: DNA is fragmented and short adapters are added to both ends, often with index sequences (barcodes) so many samples can share one run.
2. Cluster generation: fragments bind to a flow cell and are copied in place into clusters of identical molecules.
3. Sequencing: fluorescent, reversibly terminated nucleotides are added one base per cycle and imaged, so every cluster is read in parallel.
4. The output is short reads (commonly 50 to 300 bases, often paired-end) in FASTQA file format that stores each sequencing read with a quality score for every base. files, which store each read with a quality score per base.

Long-read technologies (Pacific Biosciences and Oxford Nanopore) read single molecules thousands to over a hundred thousand bases long. They help with repeats, structural variants and assembling new genomes.

Key numbers you will see

• Phred qualityA score Q = -10 log10(error probability); Q30 means a 1 in 1,000 chance a base is wrong. (Q): Q = -10 log10(error probability). Q20 means a 1 in 100 chance a base is wrong; Q30 means 1 in 1,000, a common quality benchmark.
• CoverageHow many reads cover each base on average: number of reads times read length divided by genome size. (depth): how many reads cover each base on average.

coverage = number of reads x read length / genome size

Worked example. 40 million reads of 150 bases on a 5 million base bacterial genome: 40,000,000 x 150 / 5,000,000 = 1,200x coverage, far more than the roughly 30 to 100x usually wanted. You could multiplex many more samples on the same run.

Choosing a method

Clinical sequencing tests are validated and run under laboratory regulations and accreditation (in the US, CLIA and often CAP), and results are reviewed by qualified staff. Sample handling happens in the lab under supervision.

Common mistakes

• Sending a mixed template (two PCR products) for Sanger and getting an unreadable trace.
• Expecting Sanger to detect a variant present in 5 percent of cells.
• Ignoring the first and last bases of a Sanger read, where quality is usually poor.
• Planning NGS without calculating the coverage you need.
• Mixing up sample indexes, which assigns reads to the wrong sample.

Quick recap

• Sanger: ddNTP chain termination plus capillary electrophoresis; one region, very accurate, about 1 kb.
• NGS: library prep, clusters, sequencing by synthesis; millions of short reads in FASTQ.
• Long reads help with repeats and structural variants.
• Know Phred scores (Q30 = 1 in 1,000) and coverage = reads x length / genome size.

Try it

The Do tab has a quiz on choosing and interpreting sequencing methods.