Types of PCR: Which One Actually Fits Your Experiment?

You Do Not Need Fifteen PCRs

The literature names at least fifteen PCR variants — hot-start, touchdown, long-range, RACE, methylation-specific, asymmetric, and more. Most are narrow solutions to a specific annoyance, not tools a working lab reaches for daily. In this guide I’ll walk through the six or seven you will actually use in a molecular biology lab, organised not as a list but as a decision tree: your starting material, whether you need a number, and how many targets you’re chasing. The goal is that when your supervisor says “just run PCR,” you know which PCR they mean.

Pick a Positive Control as a Stand-In

Before we choose between types, a practical habit. Whatever PCR type you land on, I keep one validated control pair as an internal yardstick so I can tell a technique problem from a reagent problem. The GAPDH pair at OriGene HP205798 — forward GTCTCCTCTGACTTCAACAGCG, reverse ACCACCCTGTTGCTGTAGCCAA, 131 bp amplicon, Tm 61.6°C / 65.1°C, annealing 58.3°C — is my daily baseline. Cheap, works in conventional, qPCR, and (as cDNA) RT-qPCR alike. It also means a failing gel points at my reagents rather than my primers.

The Decision Framework

Do not memorise the types. Memorise the decision tree instead:

Start | your experiment
  |
  +-- Is your starting material RNA? ----------> RT step needed
  |       +-- Want a number? -----------------> RT-qPCR
  |       +-- Want yes/no (presence)? --------> RT-PCR
  |
  +-- Need a number (quantification)?
  |       +-- >2-fold changes, cheap ---------> qPCR
  |       +-- Rare allele / absolute copies --> digital PCR
  |
  +-- Need multi-target detection? -----------> Multiplex PCR
  |
  +-- Need maximum specificity on dirty DNA? -> Nested or hot-start
  |
  +-- Presence/absence or cloning? -----------> Conventional PCR

Four branches, each driven by a single, answerable question. Each pick buys you — and what it costs.

Conventional (Endpoint) PCR

What it does: Amplifies a target, then you read the product on an agarose gel — a band or no band.

When to use it: Presence-absence questions, genotyping (including SNP PCR for it tests), cloning, colony screening, and confirming a restriction digest.

When NOT to use it: It cannot quantify — the gel endpoint has saturated by the final cycles, so band intensity is not a concentration. If your experiment needs a number or a fold-change, choose otherwise. It also cannot tell you about early cycles, so any real-time dynamic is invisible at saturation.

Quantitative PCR (qPCR)

What it does: monitors fluorescence every cycle and compares a threshold cycle (Cq) against a standard curve or reference gene, giving relative quantification over the exponential phase.

When to use it: gene expression, pathogen detection with a number, GMO and clinical assays, any >2× gene change.

When NOT to use it: SYBR Green binds any double-stranded product — it isn’t allele-specific, and primers dimers inflate the signal. It needs a standard curve for absolute numbers, and below about ten copies its coefficient of variation climbs past 20%. Use a melt curve or probe chemistry to confirm specificity.

Reverse Transcription PCR (RT-PCR) and RT-qPCR

What it does: RNA → cDNA via reverse transcriptase, then PCR. RT-PCR alone gives presence-absence from RNA; RT-qPCR quantifies it.

When to use it: any time your target is RNA — gene expression, RNA-virus detection (including SARS-CoV-2), transcriptome work. Choose between one-step (single reaction, RNA in, cDNA plus PCR in one tube — fewer handling steps, fewer chances to contaminate) and two-step (separate RT then qPCR — separate reactions and you can use the same cDNA for many targets).

When NOT to use it: forgetting the reverse-transcription step — a DNA polymerase cannot use RNA as template, so you get nothing on the gel. And do not confuse it with real-time: RT = reverse transcription, qPCR = quantification; they are different axes. The naming trap ate many a new lab. (See RT-PCR vs qPCR .)

Nested PCR

What it does: two sequential rounds, outer primers first, then inner primers amplifying a region inside the first product, for extra specificity and sensitivity.

When to use it: degraded DNA (ancient, forensic, environmental), low-copy targets, or when the primer design is intrinsically tricky.

When NOT to use it: each round means opening the tube — the classic contamination channel. Nearly every diagnostic lab avoids nested tube-to-tube precisely because carryover of the first-round product fakes positive results. Use closed-tube variant or a physical separation if you must, and never read a nested gel without a no-template control.

Multiplex PCR

What it does: multiple primer pairs in one reaction to amplify several targets simultaneously.

When to use it: forensic STR panels, pathogen panels, microsatellite and SNP screening, anything where sample is short.

When NOT to use it: a multiplex reaction drops one quiet target because two primers cross-dimerised and it was out-competed — or the stoichiometry of the highest primer concentration hushes the lower ones. Designing compatible Tm and amplicon size across all pairs is a real job; without careful dimer checking you get phantom amplicons or dropouts. Either fix it by careful design or run simplex.

Digital PCR (dPCR)

What it does: partitions the sample into thousands of nano/pico-litre reactions, runs to endpoint, then counts which wells are positive and applies Poisson stats — giving copies per microliter with no standard curve.

When to use it: rare-allele detection in liquid biopsy (down to ~0.1% and below), copy-number on small fold-changes, absolute reference standards, detecting sub-fold-copy number and low-copy pathogens. Digital stays precise (CV under ~10%) at only a few copies per reaction, where qPCR CV climbs past 20% below ~10 copies.

When NOT to use it: if you’re just measuring a healthy >2× expression change across a hundred samples, the extra cost per reaction and the slow throughput buy you almost nothing. And in a 384-well plate workflow it is impractical. Reserve it for the rare-allele and absolute-quantity jobs qPCR genuinely cannot do.

The Not-a-Type Type: Hot-Start Polymerase

Hot-start is less a PCR type than a modification you apply to the others: a polymerase that is dead until the initial denaturation kills primer-dimers formed during setup (room temperature). The extra cost is modest and the benefit — fewer spurious bands — makes it my default polymerase for endpoint and multiplex reactions. It will not fix a mis-anneal, but it removes the most common setup artefact.

Worked Decision: A Post-BRCA1-Editing Check

Suppose I want to know whether a CRISPR edit knocked down the BRCA1 transcript level. My starting material is RNA. I need a number to compare edited vs control — not just “faint band on a gel.”

Decision tree: RNA → RT branch; quantify → RT-qPCR. I go two-step (RT to cDNA once, use that cDNA across several BRCA1 and GAPDH-control wells). It’s mRNA, so the amplicon stays under 150 bp and it’s made against the spliced read. Two weeks later if the same supervisor asks me to verify a plasmid construct is present in transformed colonies, I’m now asking a DNA presence/absence question — so it falls back to conventional PCR on a gel, the exact opposite branch. Same two words “run PCR,” radically different types.

Quick-Reference Table

How I Pick a PCR Type in Five Minutes

When a protocol says \u201cPCR\u201d and nothing else, I run a quick mental filter before touching a pipette:

1. Template first. RNA? Then the RT branch is non-negotiable; DNA keeps the rest of the tree.
2. Question second. Presence-absence vs a number changes everything \u2014 band vs Cq vs droplet count.
3. Precision third. Am I resolving a 1.3\u00d7 change or a 50\u00d7 change? The smaller the change, the more I lean toward dPCR or the tightest qPCR controls.
4. Throughput last. A hundred samples to screen favours multiplex or 384-well qPCR; one rare-allele question favours dPCR even at higher cost.

That order \u2014 template, question, precision, throughput \u2014 has steered me right more often than any single list of \u201cbest\u201d PCRs. Once the type is chosen, primer design follows: short amplicons for qPCR, Tm-harmonised pairs for multiplex, two non-overlapping sets for nested. The type determines the design constraints, and getting that first decision right is the cheapest step of the experiment.