Assay Development and Design

Working backward from the result

An assay turns the presence, amount, or sequence of a nucleic acid into a readable result. Design starts from the answer you need and how it will be read, then chooses a target, reagents, conditions, and controls to get there ¹. Proving the finished assay works - verification and validation - is a separate step, covered in the lab operations course.

Start with the target

“Detect the virus” isn’t a target. A target is a specific sequence:

• DNA or RNA? RNA targets need reverse transcription first.
• Which region? One present in everything the assay must catch.
• What makes a positive? Presence of a sequence (is Chlamydia there?) is a different design problem from a single base change (is this Factor V Leiden?).

The region must be conserved across everything the assay should detect (inclusivity) and different from everything it should ignore (exclusivity) - near relatives, normal flora, the human genome ¹. Designers align many sequences and look for stretches where all targets agree and neighbors don’t.

Targets drift. Viral escape from primers is real: some SARS-CoV-2 variants carried a deletion that knocked out one S-gene target in a widely used assay (S-gene target failure), which labs then used as a variant marker. That’s one reason many assays hit two targets: the CDC’s 2019-nCoV panel used two nucleocapsid-gene regions, N1 and N2.

Primers, probe, and amplicon

Primers, probe, enzyme, and buffer are designed as one set, since each constrains the others. Typical rules for a probe-based real-time PCR:

The 3’ end is critical: polymerase extends from the 3’-OH, so a mismatch there blocks extension, while a mismatch near the 5’ end is often tolerated. Allele-specific PCR exploits this by putting the variant base at a primer’s 3’ end.

Enzyme and buffer follow the goal (see Polymerases): a Taq-type hot-start enzyme for hydrolysis probes; Mg²⁺ tuned for specificity ²; a one-step RT-PCR mix for RNA targets.

The goal shapes the design

The same target can serve all three; the goal changes the controls, standards, and reporting.

Analytical sensitivity and specificity

• Analytical sensitivity is how little target the assay can detect, expressed as the limit of detection (LoD): the lowest concentration detected in about 95% of replicates ¹. Short amplicons, efficient extraction, multicopy targets, and larger input volumes all help.
• Analytical specificity is responding only to the intended target. Failure is cross-reactivity with related organisms or alleles, or interference from substances in the specimen (heme, heparin, mucus) that inhibit the reaction ¹.

These differ from clinical sensitivity and specificity, which ask how well the test classifies patients.

Controls

Controls are designed in from the start, and each answers one question:

A worked case: in the CDC SARS-CoV-2 panel, each specimen was tested for N1, N2, and human RNase P (RP). RP is a human gene, so a positive RP in a sample negative for N1 and N2 shows that human cells were collected and extraction and RT-PCR worked - a valid negative. No RP and no N1/N2 means the result is invalid, not negative. Many commercial assays instead spike a non-human exogenous internal control (such as an engineered RNA or phage) into each sample before extraction.

Readout and interpretation

The chemistry exists to produce something a detector and a rule can read ¹:

• A fluorescent probe signal needs a probe and an enzyme activity designed together (TaqMan needs Taq’s 5’→3’ nuclease).
• A fragment-size readout needs fragments the gel or capillary can resolve.
• A quantitative result needs calibrators and a defined reporting unit.
• Every result needs a cutoff: what Ct counts as positive, what to do with a late or single-target result.

A design that makes molecules nobody can read, or signals no rule can interpret, has failed no matter how good the chemistry is.

What to take from this

Start from the question and readout. Choose a target region conserved in everything you must catch and different in everything you must ignore, ideally two targets. Fit 18-25 nt primers (Tm about 60 °C, no 3’ dimers) and a probe about 8-10 °C hotter inside a 70-150 bp amplicon, remembering that 3’ mismatches block extension. Qualitative, quantitative, and genotyping goals carry different burdens. Analytical sensitivity is the LoD; specificity is the absence of cross-reactivity and interference. Positive, negative, no-template, internal, and extraction controls each answer one failure question; RNase P in the CDC panel is the model internal control.