Setting Threshold and Baseline in qPCR: How Misplacement Distorts Your Fold Changes
Threshold and baseline settings in qPCR analysis directly determine your Ct (Cq) values—and by extension, every fold-change calculation downstream. A threshold set too high pushes Ct values into the plateau phase where amplification is no longer exponential. A threshold set too low sits in the noise. Either way, your reported gene expression changes will be wrong, and the error is invisible unless you know what to look for.
This guide explains how to set the threshold and baseline correctly for qPCR analysis, covering both automatic and manual approaches, with a worked example showing how threshold misplacement distorts your results.
What the Baseline and Threshold Actually Measure
Every qPCR amplification curve has three phases:
- Baseline (background) phase — cycles 1 through ~15, where fluorescence is indistinguishable from instrument noise and non-specific background signal
- Exponential phase — cycles where the target is doubling (or nearly doubling) each cycle. This is where quantification must happen.
- Plateau phase — amplification slows as reagents are consumed and product accumulates. Fluorescence differences between samples narrow and quantification becomes unreliable.
The baseline defines what counts as “no signal.” The software subtracts this background from all fluorescence readings. The threshold is a horizontal line drawn across the amplification plot. The cycle where each sample’s curve crosses the threshold is its Ct (or Cq) value.
MIQE guidelines recommend using “Cq” (quantification cycle) as the standard term. Instrument software may display “Ct” (Applied Biosystems/Bio-Rad) or “Cp” (Roche). These are functionally equivalent—the cycle number at which fluorescence crosses the defined threshold.
How to Set the Baseline: Step by Step
Incorrect baseline settings are the most underappreciated source of Ct error. If the baseline window includes cycles where early amplification has already started, the software subtracts real signal as background, artificially lowering fluorescence and inflating Ct values.
Step 1: View the Amplification Plot in Linear Scale
Switch from logarithmic to linear fluorescence scale. In linear view, the baseline phase is the flat region at the bottom of the plot. In log scale, early amplification is visible but the baseline/exponential boundary is harder to identify precisely.
Step 2: Identify Where the Earliest Sample Begins to Rise
Find the sample with the lowest Ct (highest abundance). This sample will show exponential rise first—typically between cycles 10 and 20 for moderate-abundance targets. Note the cycle where its fluorescence first visibly separates from the flat baseline.
Step 3: Set the Baseline End Cycle
Set the baseline end at least 2 cycles before the earliest sample begins to rise. For example, if your most abundant sample begins exponential rise at cycle 14, set the baseline end to cycle 12 or earlier.
Leaving the baseline end at the instrument default (often cycle 15 or “auto”) when you have high-abundance samples that begin amplifying at cycle 12–13. The software subtracts real amplification signal as background, adding 0.5–1.5 cycles of error to those samples.
Step 4: Set the Baseline Start Cycle
The baseline start should be cycle 3 or later. Cycles 1–2 often have unstable fluorescence from initial temperature equilibration. A typical baseline window is cycles 3–12, but adjust based on your data.
Step 5: Apply Per-Target, Not Per-Plate
Set the baseline independently for each target gene on the plate. A reference gene amplifying at cycle 15 needs a different baseline window than a low-abundance target amplifying at cycle 28. Most instrument software allows per-detector or per-target baseline settings.
How to Set the Threshold: Manual Placement
The threshold must sit in the exponential phase of all amplification curves for the target being analyzed. This is the region where relative fluorescence differences between samples faithfully represent starting template differences.
Rule 1: Place the Threshold in the Exponential Phase
On a logarithmic amplification plot, the exponential phase appears as a straight, steep region. Place the threshold where the curves are parallel and linear on the log scale. Avoid the early shoulder (where curves are just leaving baseline) and the upper bend (where curves approach plateau).
Rule 2: All Curves Must Cross the Threshold in Their Exponential Phase
If you have samples ranging from Ct 15 to Ct 32, verify that the threshold intersects each curve during its exponential phase. High-Ct samples with low template may have shorter exponential regions, requiring a lower threshold to catch them before they plateau.
Rule 3: The Threshold Must Be the Same for All Samples of the Same Target
When comparing treated vs. control samples for the same gene, they must share the same threshold value. Moving the threshold between samples invalidates any delta delta Ct calculation because you are comparing Ct values measured at different fluorescence levels.
Rule 4: The Threshold Must Be Above Background Noise
The default automatic threshold in many instruments is set at 10 standard deviations above the mean baseline fluorescence. This is a reasonable starting point. If you see Ct values assigned during the baseline phase (fluorescence clearly still flat), your threshold is too low.
Automatic vs. Manual Threshold: When to Override
Modern instrument software (Bio-Rad CFX Maestro, Applied Biosystems QuantStudio, Roche LightCycler) includes automatic threshold algorithms. A 2021 review in Clinical Chemistry highlighted how automatic settings can introduce systematic bias across platforms. These work well for clean data but can fail in specific scenarios:
| Scenario | Auto Threshold Behavior | Action |
|---|---|---|
| Clean data, similar Ct range | Works well | Accept auto |
| Wide Ct range (>15 cycles between samples) | May place threshold too high for low-abundance samples | Lower manually |
| Noisy baseline (degraded reagents) | Inflated baseline raises threshold into plateau | Adjust baseline first, then re-auto or set manually |
| Very low abundance targets (Ct >33) | May not detect crossing | Lower threshold; verify against NTC |
| Multi-plate experiment | Different auto threshold per plate | Set one manual threshold across all plates for comparability |
If your experiment spans multiple plates, use the same manual threshold value for each target gene across all plates. Automatic thresholds are calculated per plate and will differ between runs, introducing systematic plate-to-plate variation into your Ct values.
Worked Example: How Threshold Placement Changes Your Fold Change
Consider a simple experiment: one target gene, one reference gene (GAPDH), two conditions (control and treated), three biological replicates each.
True Ct values (threshold correctly placed in exponential phase):
- Target gene, control: Ct = 22.0 (mean of 3 replicates)
- Target gene, treated: Ct = 24.5
- GAPDH, control: Ct = 18.0
- GAPDH, treated: Ct = 18.0
Delta Ct (control) = 22.0 − 18.0 = 4.0
Delta Ct (treated) = 24.5 − 18.0 = 6.5
DDCt = 6.5 − 4.0 = 2.5
Fold change = 2^(−2.5) = 0.177, i.e., the target is 5.7-fold downregulated.
Now suppose the threshold is placed too high—in the early plateau rather than the exponential phase. The high-Ct samples (treated) are affected more because their exponential region is narrower:
Observed Ct values with threshold in early plateau:
- Target gene, control: Ct = 23.2 (shifted +1.2 cycles)
- Target gene, treated: Ct = 27.0 (shifted +2.5 cycles—hit plateau earlier)
- GAPDH, control: Ct = 19.0 (shifted +1.0)
- GAPDH, treated: Ct = 19.1 (shifted +1.1)
Delta Ct (control) = 23.2 − 19.0 = 4.2
Delta Ct (treated) = 27.0 − 19.1 = 7.9
DDCt = 7.9 − 4.2 = 3.7
Fold change = 2^(−3.7) = 0.077, i.e., the target appears 13-fold downregulated.
The true fold change was 5.7-fold downregulation. The misplaced threshold reports 13-fold—a 2.3-fold exaggeration of the effect. This distortion is not random; it systematically inflates differences because low-abundance samples are disproportionately affected by threshold placement in the plateau.
Checking Your Settings: A Post-Analysis Audit
After setting baseline and threshold, verify your configuration before exporting Ct values:
- Baseline check: View the baseline-subtracted amplification plot. No sample should show negative fluorescence in the baseline region. Negative values mean your baseline is overcorrecting.
- Threshold position: On a log-scale amplification plot, the threshold should cross all curves in their linear (parallel) region. If any curve crosses the threshold at a bend, adjust.
- NTC check: Your no-template controls should either not cross the threshold, or cross it at least 5–7 cycles after your lowest-abundance sample.
- Replicate consistency: Technical replicates should be within 0.5 Ct of each other for Ct values under 30, and within 1.0 Ct for values between 30–35. Wider variation suggests baseline or threshold issues (or pipetting errors).
If your replicates for a given sample scatter more than expected, check whether the threshold intersects their curves at different phases (one in exponential, one in plateau). This is a telltale sign of threshold misplacement.
Connecting to Efficiency and Normalization
Threshold and baseline settings affect more than individual Ct values. They cascade into your efficiency calculations from standard curves—if the threshold crosses your dilution series at inconsistent phases, the slope (and therefore efficiency) will be wrong. They also affect reference gene stability analysis, since GeNorm and NormFinder M-values are computed from Ct differences that depend on accurate threshold placement.
Getting threshold and baseline right is foundational. Every downstream calculation trusts that your Ct values represent the same point in each sample’s amplification curve. Verify this before moving on to analysis.