Next-Generation Sequencing

Millions of reads at once

Sanger reads one fragment at a time. Next-generation sequencing reads millions to billions of short fragments in parallel, then lines them up by computer ¹.

Library preparation

1. Fragment DNA to a few hundred bases (enzymatic or acoustic shearing); cell-free DNA is already about 170 bp.
2. Repair the ends and add an A overhang.
3. Ligate adapters: known sequences that let fragments bind the flow cell and be primed for sequencing ¹.
4. Indexes (barcodes) in the adapters label each patient’s library, so dozens of samples can be pooled in one run and sorted out afterward (demultiplexing).
5. Optional unique molecular identifiers (UMIs): random tags on each original molecule (see Reads and depth).
6. Amplify, quantify (fluorometry or qPCR), and pool in equal amounts.

Target enrichment

¹

Whole-genome sequencing skips enrichment entirely.

Clonal amplification

One molecule is too faint to see, so each is copied into a local cluster of identical copies ¹:

• Bridge amplification: fragments bind oligos on a flow cell and copy themselves into neighboring clusters.
• Emulsion PCR: each fragment is amplified on its own bead in an oil droplet.

Sequencing by synthesis

A polymerase extends a primer on every cluster at once, one base per cycle ¹:

• Fluorescent reversible terminators: each nucleotide carries a dye and a block; one base adds, the flow cell is imaged, then dye and block are removed. The cluster’s color each cycle gives its sequence.
• Semiconductor detection: nucleotides are flowed one type at a time; incorporation releases a hydrogen ion, sensed as a pH change on a chip. Homopolymer runs (AAAAA) are this method’s weak point, because signal must be counted by size.

Both build the complementary strand and watch incorporation, the same logic as replication ².

Reads and depth

• Read length: typically 75-300 bases per read.
• Paired-end: reading both ends of each fragment improves alignment and detects insertions and rearrangements.
• Depth (coverage): how many reads cover a base. Germline tests often aim for 30x or more; tumor panels several hundred to thousands, because variants may sit in a small fraction of cells.
• Duplicates: PCR copies of one original molecule add reads without adding information; they’re marked and removed.
• UMIs: reads sharing a UMI come from one original molecule. A true variant appears in all of them; a PCR or sequencing error usually in just one. Collapsing families lets liquid-biopsy assays detect variants well below 1%.

Choosing a test

Panels give depth on a few genes; exomes and genomes give breadth, with more data to interpret and more incidental findings to manage.

What to take from this

NGS fragments DNA, ligates adapters with patient indexes (and optionally UMIs), enriches targets by hybrid capture or amplicon PCR, clonally amplifies each fragment, and reads millions of clusters by synthesis. Depth is reads per base - 30x for germline, hundreds to thousands for tumors - and UMIs collapse duplicates for very low-level variants. Panels answer defined questions deeply; exomes and genomes cast the wide net.