coverage / depth

Imagine proofreading a page by getting many people to each read a few overlapping lines. If every spot on the page has been read by, say, thirty different people, you can trust the consensus even if any one reader makes a slip. Sequencing coverage (or depth) is exactly this idea for DNA: how many separate reads cover each position in the genome.

Because individual sequencing reads are short and somewhat error-prone, a sequencing experiment deliberately reads the same region over and over from many different fragments. Coverage at a given position is the number of reads that overlap it; the overall depth, written like 30x, means on average each base was read about thirty times. Higher depth means more independent looks at each base, so genuine differences from the reference stand out clearly while random read errors get outvoted by the majority. Coverage is usually uneven across a genome — some regions, especially repetitive or GC-extreme ones, get fewer reads and so are read less reliably.

Depth is one of the most important quality dials in sequencing, and it directly shapes what you can conclude. Calling a heterozygous variant (where one of your two gene copies differs) confidently needs enough depth to see both versions; detecting a rare mutant present in only a few percent of cells (as in a tumour) needs very high depth. But more is not free: depth costs sequencing capacity and money, so experiments are designed to hit a target depth appropriate to the question. A claim of a detected mutation backed by only one or two reads at low depth should be treated with healthy skepticism.