digital PCR

Suppose you want to count exactly how many target molecules are in a sample — not 'roughly twice as much as the other tube,' but an actual number. Quantitative PCR struggles with this because it reads brightness against a curve. Digital PCR takes a cleverer route: instead of measuring one bulk reaction, it splits the sample into thousands of tiny separate reactions and simply counts how many of them light up.

The sample is partitioned, often into thousands or millions of microscopic water droplets in oil or wells on a chip, so that on average each partition holds either zero or one target molecule. Then PCR runs in every partition at once. A partition that contained a target molecule glows positive after amplification; an empty one stays dark. You count the positives. Because the molecules were distributed essentially at random, a bit of statistics (the Poisson distribution, which accounts for the few partitions that happened to get two molecules) converts the fraction of glowing partitions into an absolute count of starting molecules — no standard curve needed.

Digital PCR shines where qPCR is weak: it gives absolute counts, it is more precise for tiny differences, and it tolerates inhibitors better, because each partition is its own clean reaction. That precision matters for spotting a rare mutant DNA molecule hiding among thousands of normal ones — for example a few tumour DNA fragments in a blood sample (a liquid biopsy) — or for measuring exact copy numbers. The trade-offs are higher cost and specialized instruments, so digital PCR is reserved for jobs that genuinely need its accuracy rather than as an everyday replacement for qPCR.