Digital PCR

Turns partition counts into absolute copies per reaction with a confidence interval, and shows the ratios between targets without needing a standard curve.

What it doesCopy Anchor Link

Digital PCR does not measure a cycle. The reaction is split into many small partitions, each either containing template or not, and after amplification you count how many came up positive. Because partitions are loaded at random, the fraction that stay negative is enough to recover the average number of template molecules per partition — no standard curve, no efficiency assumption, no reference gene.

The counting is Poisson, and that matters in both directions. At low occupancy nearly every positive partition is one molecule and the count is almost the answer. As occupancy rises, partitions start receiving two or more, positives saturate, and the correction that recovers the true concentration grows steeply — which is also where the uncertainty grows. This reports both the estimate and that uncertainty rather than a bare copy number.

When you'd use thisCopy Anchor Link

• When you need an absolute concentration, not a relative one — quantifying a library before sequencing, calibrating a standard, certifying a reference material.
• For rare targets. Counting positive partitions is a more forgiving measurement of a few molecules than reading a late Cq off a shallow curve.
• For small fold differences. Allelic ratios and copy-number changes near 1.5× are where relative quantification runs out of resolution and counting does not.
• In inhibitor-rich samples, where an efficiency assumption is exactly the thing you cannot make.

What this computesCopy Anchor Link

• Copies per partition from the negative fraction, by the Poisson relationship — the estimate the rest is derived from.
• Copies per microlitre of reaction, using the partition volume and the reaction and template volumes you enter, each of which is a control rather than a built-in constant.
• A confidence interval on that concentration at a level you set, propagated from the partition counts.
• The ratio between targets, with its own interval — the quantity a copy-number or allelic question actually asks about, which is not the same as dividing two point estimates.
• The partition accounting per well — positives, negatives and total — so a well that lost partitions is visible rather than silently folded into the total.

Getting the counts inCopy Anchor Link

Partition counts arrive one of three ways, and none of them asks you to retype a number.

• Paste the table your instrument software exports: one row per well, with the positive and accepted partition counts. The column names instruments actually print are accepted as they are — Bio-Rad's AcceptedDroplets, QIAcuity's Valid partitions — so an export needs no renaming.
• Open the run file. RDML is the open qPCR exchange standard, and from version 1.3 it carries digital data: per-well positive and negative counts, and the average volume of one partition, which is the number the concentration is divided by. Both travel into the panel together, and the volume the file states replaces the one in the box rather than sitting silently beside it. A run that declares two different partition volumes sets neither and says so — averaging them would invent a number no instrument reported.
• Start from the droplets themselves if you have amplitudes rather than counts, on the Droplets and Classify tabs, which do the classifying and hand the counts here.

A digital run's plate is drawn like any other: every well coloured by its own concentration, so a dilution series reads as a gradient across the plate and a dead row reads as a dead row. The colour runs on a logarithmic scale, which is what makes that gradient readable: a dilution series covers orders of magnitude by design, and spaced linearly every well below the top decade would be the same pale cell. The legend says so, and the exact number is in each cell and in its tooltip either way.

A well where every partition came up positive is left uncoloured rather than clamped — its concentration is above the assay's range, and a colour would imply a number. A well where none came up positive is a real reading of zero, not a missing one, so it sits at the pale end of the scale and is counted underneath the plate: zero has no place on a logarithmic axis, and quietly moving it onto one would flatten every well above it.

You do not have to open the panel to look at a run. An .rdml in a project opens on its own, like a sequence or a trace: the plate, and the raw amplification and melting curves exactly as the instrument wrote them — no baseline correction, no normalisation, no threshold. Every run in a multi-run file is there, and a view the run holds no data for says which rather than drawing an empty axis. Nothing is analysed in that view; it is for seeing what a file contains before deciding what to ask of it.

AutomationCopy Anchor Link

Available as the quantify_dpcr workflow node and to the copilot, with the same controls this panel has: the partition volume and reaction geometry, the dilution chain, which of the eleven confidence intervals is carried onto the concentration scale, whether replicate wells are pooled, and — added 2026-09-03 — the target-vs-reference copy number with the reference ploidy.

Reading the run file is automated too, as read_rdml_run: it lists what every run in a file holds — curves, melting curves, instrument Cq values, partition counts and the partition volume — before importing anything, then hands back whichever table you ask for, in the same form the panel uses.

The copy number is reported only when you name both assays. Left unnamed it returns nothing rather than pairing two assays it guessed at, because a guessed pair is a comparison nobody asked for.

What it does not doCopy Anchor Link

It assumes partitions are equal in volume and independently loaded. It cannot see a partition that failed to amplify for a reason other than being empty, which is why the raw positive and negative counts stay on screen next to the derived concentration instead of being replaced by it.