How to design a ddPCR assay in silico

What a ddPCR assay needs

A standard ddPCR assay is a hydrolysis-probe (TaqMan-style) assay: a forward primer, a reverse primer, and a fluorescent probe that binds between them. ddPCR partitions the reaction into thousands of droplets and counts positive droplets, so the design goals differ slightly from qPCR: a short amplicon and a clean, specific probe signal matter more than a wide dynamic range.

Step 1: Select the target region

Choose the region that contains what you want to measure: a gene, a transcript junction, or a specific variant such as a point mutation or a copy-number target. If you are detecting a mutation, center the design on the variant position, because the probe usually needs to sit over it. Pull the reference sequence from a primary database (for example NCBI or Ensembl) and record the exact coordinates so the design is reproducible.

Step 2: Design the primers

Design the two primers with these typical starting targets, then adjust for the specific sequence:

• Melting temperature (Tm) around 58 to 62°C, with the two primers matched to within about 2°C of each other
• Length roughly 18 to 24 nucleotides
• GC content around 40 to 60%
• A short amplicon, commonly 60 to 150 bp, which suits ddPCR droplet chemistry
• Avoid long single-base runs and strong 3′ self-complementarity that cause primer dimers

Step 3: Design the probe

The hydrolysis probe carries a fluorophore at the 5′ end and a quencher at the 3′ end, and binds between the primers on whichever strand gives the cleaner sequence:

• Tm about 5 to 10°C higher than the primers, so the probe binds first
• Avoid a G at the 5′ end, because it quenches the fluorophore even after cleavage
• For allele discrimination (mutant vs wild type), place the probe directly over the variant, or use two spectrally distinct probes, one per allele
• Keep it as short as the Tm target allows; modified bases such as LNA or an MGB group help reach the Tm on short probes

Step 4: Check specificity in silico

This is the step a chatbot cannot do, and where most bad assays are caught. Before ordering anything, verify the design computationally:

• Run in silico PCR against the target genome to confirm a single correct amplicon and no off-target products
• BLAST each primer and the probe to check for unintended binding sites elsewhere in the genome
• Check for common variants (SNPs) under the primer and probe binding sites using dbSNP or gnomAD; a SNP under a primer can silently drop an allele
• For multiplex or closely related targets, confirm the primers and probes do not cross-react

Step 5: Multiplexing (optional)

ddPCR often runs several targets in one reaction. To multiplex, give each target a probe with a spectrally distinct fluorophore, keep all primers at a similar Tm so they amplify together, and confirm that no primer or probe from one assay binds a target from another. Amplitude-based multiplexing (varying probe concentration) can add more targets per channel, but start with distinct fluorophores.

Common pitfalls

• A SNP under a primer or probe binding site, which can cause allele dropout
• An amplicon that is too long, which lowers droplet efficiency
• A probe Tm too close to the primer Tm, which weakens the signal
• Secondary structure or primer dimers that were never checked
• Off-target amplification from primers that were not BLAST-checked

A worked example

A concrete illustration with typical numbers: to detect a point mutation you might design primers of about 20 nucleotides with melting temperatures near 60°C (matched within roughly 2°C), an amplicon of about 90 bp, and a hydrolysis probe of about 15 nucleotides at a Tm near 68°C placed over the variant. In silico PCR should then return a single 90 bp product, BLAST should show no close off-target binding, and a dbSNP check should confirm no common variant sits under either primer. Only then is the assay ordered.

For a real, documented ddPCR design that detects a cancer hotspot variant end to end, with the specificity checks and a full reproducibility folder, see the NSCLC hotspot ddPCR case study. For the broader idea of running this design work autonomously, see What is an AI in silico designer for molecular biology?.