Comparing RT-qPCR and RNA-seq on the same question

Point RT-qPCR and RNA-seq at the same gene name and you can still be measuring two different molecules. One method is a primer's opinion about a short amplicon. The other is a count over whatever the annotation grouped under that gene. This explainer is for the moment those answers disagree, and for the earlier moment when you choose which method should carry the question. Both sit on the path in from cells to a gene expression result.

Same gene name, different measurement

RNA-seq is a discovery instrument when the gene list is not yet closed. You pay in depth, in replicates and in analysis for a table of many features. RT-qPCR is a confirmation and a measurement instrument when the transcripts are already named. You pay in primer design and in reference-gene assumptions for a precise relative comparison of a short list. Using each method for the other's job is why people call one of them broken.

Dynamic range is not a single trophy. A qPCR assay with a linear dilution series can follow one target across many doubling intervals, down to a few molecules in the well, where Poisson noise and contamination become the limit. Very high template becomes hard for a different reason: the signal rises inside the baseline window. RNA-seq's range depends on how many reads the library received and on what else was in the RNA. A rare transcript can be zero in the count table while a qPCR of the same RNA still crosses threshold. An extremely abundant transcript can dominate the library and shrink every other gene's share. Neither fact makes one method the more accurate one. It makes them sensitive to different failures.

The MIQE guidelines say what a qPCR comparison should report. A sequencing technology overview from Illumina is the matching background for the count side. Neither document requires the two numbers to be equal.

What each method is physically counting

Reverse transcription copies RNA into cDNA. In RT-qPCR, two primers then define a short product. If both primers sit in an exon shared by every isoform, the assay is closer to a gene-level measurement, and it is still blind to anything outside that amplicon. If a primer sits on an exon-exon junction, the assay is isoform-minded, and it will miss a splice form that skips the junction. Genomic DNA will interfere according to how the primers were placed. The working habits for the relative calculation are in RT-qPCR for relative expression.

RNA-seq fragments the RNA, or the cDNA, and counts pieces that a pipeline assigns to a feature. A gene-level count adds the isoforms the annotation links. A transcript-level count tries to split them and is only as good as the annotation and the depth. A primer that interrogates isoform A and a table that reports A plus B plus C are allowed to disagree. They answered different questions that happened to share a gene symbol.

Length adds another split. RNA-seq quantification is affected by transcript length because longer transcripts yield more fragments. TPM-style scalings try to correct that within a sample. A qPCR amplicon is the same length in every sample, so length is not the variable. A gene that looks unchanged by qPCR and changed by a length-naive count, or the reverse, may be an annotation and scaling issue rather than a wet-lab failure.

Assay classes on one bench

qPCR reagents are a master mix, primers, and a reverse transcriptase, in the classes stocked as molecular biology. RNA-seq reagents are a library kit and a sequencing run. The bridge between them is the RNA itself. If the RNA for the qPCR was a different extraction from the RNA that was sequenced, you are not comparing methods on one question. You are comparing two experiments.

Batch sits in that gap. A qPCR plate has its own day effect. A sequencing lane has another. If all the treated samples were on one plate and one lane, and all the controls on another, both technologies will agree and both will be wrong about the cause. Interleave the biological groups inside each technology. Agreement between two confounded assays is not reassurance.

A comparison that is allowed to disagree

Decide the job before the data exist. Discovery of which transcripts move is an RNA-seq job, followed, if you need it, by qPCR on independent samples for a short list you predeclare. Measuring a known transcript with a stable reference is a qPCR job, and sequencing is optional. Running RNA-seq and then qPCR on the same RNA from the same units checks the assay. It does not multiply the sample size.

When the results arrive, compare direction first, on the same contrast, after you have confirmed that the qPCR reference gene did not itself move in the count table. Then look at magnitude without demanding equality. A larger fold change in one method is not a winner. It is a prompt to check efficiency, shrinkage of log fold changes, and whether the features match.

If the direction splits, open the branch.

The primer may be on the wrong transcript. Look at the isoforms in the count table, or at a browser track, and place the amplicon on the picture. A junction the induced isoform does not contain will make qPCR look flat while the gene count rises.

Genomic DNA may be inflating the qPCR. A no-reverse-transcription well that also amplifies means the qPCR number is not RNA. The RNA-seq library, especially after a poly(A) selection or a strand-specific count, may not have treated that DNA the same way. The methods then disagree because one of them is measuring DNA. RNA handling that keeps that problem smaller is in protecting RNA during extraction.

The batches may differ. If the split appears between extraction days rather than between treatments, stop calling it a method clash.

The names may differ. A symbol in the qPCR notebook and a gene id in the table can point at neighbouring loci. Join on a stable identifier before you accuse either assay.

Where the two methods part

Disagreement with a mechanical cause

People average a qPCR fold change and an RNA-seq fold change to "meet in the middle". That average has no unit. Keep both numbers and write the feature each one measured. If you need a single claim, choose the assay whose feature matches the sentence, and use the other as a check on that feature only.

Low counts are a mechanical cause of their own. A gene with a handful of RNA-seq reads can change sign between pipelines. A Cq near the edge of detection can do the same between plates. Disagreement there is expected. It is not a reason to repeat both technologies until they coincide.

Public protocols on protocols.io show how differently groups document this check. The useful record is the primer coordinates, the annotation release, the contrast, and whether the RNA was shared. Without those, a mismatch is only a mood.

Neither result is a diagnosis

Neither method, and no combination of them, turns a research bench into a diagnostic laboratory. Biosafety follows the sample and the institution. A concordant pair of assays is still a research observation about the units you actually replicated. Do not describe agreement as clinical validation.

Local plates and shipped RNA

A common pattern is qPCR in the home laboratory and libraries sequenced elsewhere. The RNA that travels can warm in a delay, and the RNA that stays can sit in a different freezer history. Measure integrity on the aliquot that will be sequenced, not only on the aliquot you kept for qPCR. A power cut in one building and not the other is a batch even if the calendar date matches. Write the handoff: who froze the tube, when it left, and what temperature the receiving side recorded. If those histories differ, a method comparison is premature. You would be comparing a journey with a plate.

Humidity matters to the qPCR plate more than to a frozen library tube. A poorly seated optical film, the failure mode in a damp handling step, creates missing wells that have no partner in the count matrix. Do not impute them from the RNA-seq row.

What to ask when both methods are in play

State the feature, not only the gene symbol. State whether qPCR is a check on the same RNA or a measurement on new biological units. State the reference genes and whether they were examined in the count table. Sequencing can be discussed from the mRNA sequencing enquiry reference. The contrast and the table can be discussed from the differential expression analysis enquiry reference. Bring the primer coordinates to that conversation so a disagreement has somewhere to go. The pages are a discussion prompt.

The surrounding work is the nucleic acid analysis pathway. Put the shared question and the feature definition in the quote request.