Reporting Western Blot Quantification Methods for Reviewers

If a reviewer sends back "please provide details on how western blots were quantified," they're asking for about six specific things and you probably only wrote two of them. The fix is straightforward: report your detection method, how you confirmed linearity, what normalization strategy you used, how many biological replicates you ran, how you measured band intensity, and what statistical test you applied. Leave any of these out and you're inviting a revision request.

The bar for reporting has risen sharply since journals like The Journal of Biological Chemistry, EMBO Journal, and Molecular & Cellular Biology started enforcing quantification standards. What used to be acceptable — "densitometry was performed using ImageJ" — is now the equivalent of writing "statistics were done" without naming the test. Here's what a complete methods section actually looks like, piece by piece.

Detection: What Generated the Signal

State the detection chemistry and the imager. These aren't interchangeable, and they have different linear dynamic ranges that directly affect whether your quantification means anything.

• ECL + CCD camera (e.g., ChemiDoc, Azure 600, Amersham ImageQuant): Report the substrate (SuperSignal West Pico, Femto, Clarity, etc.), exposure mode (single shot vs. accumulation), and bit depth (typically 16-bit). If you used auto-exposure, say so — reviewers may flag it but at least it's transparent.
• Near-infrared fluorescence (e.g., LI-COR Odyssey CLx/M, Azure Sapphire): Report the fluorescent secondary antibodies (IRDye 680RD, 800CW, etc.), scan resolution (typically 169 µm), and intensity setting. NIR has a wider linear range (~3 orders of magnitude vs. ~1.5 for ECL on CCD), which matters for your linearity claim.
• Film: If you're still quantifying from film, report it honestly. Film has a linear dynamic range of roughly 4–8× (Gassmann et al., 2009), which means a real 10-fold change gets compressed into something that looks like 6-fold. Some journals will accept film-based quantification with caveats; others won't. Either way, state the film type, exposure time, and how you digitized (flatbed scanner model, resolution, bit depth).

The point isn't that one method is "wrong" — it's that reviewers can't evaluate your fold-change claims without knowing how the signal was captured.

Linearity: How You Know You're in Range

This is the piece most people skip, and it's the one reviewers increasingly ask about. Quantification only works within the linear dynamic range of your detection system. Outside that range, a 3-fold increase in protein looks like a 1.5-fold increase because the signal is saturated.

The gold standard is a loading series — run 2, 4, 8, 16, 20 µg of your sample on the same gel, probe, image, and plot signal vs. load. If the relationship is linear (R² > 0.98) across your working range, you're fine. Report this as: "Linearity of detection was confirmed by serial dilution of [lysate type]; signal was linear from [X] to [Y] µg total protein (R² = 0.99)."

If you didn't run a formal loading series (most people don't for every blot), you can at least state that you checked your images for pixel saturation. In Image Lab (Bio-Rad), saturated pixels show up in the "Overexposed" view. In Image Studio (LI-COR), the software flags oversaturated pixels automatically. In ImageJ/Fiji, check histogram max — if your band hits 65,535 in a 16-bit image (or 255 in 8-bit), it's clipped. Report that you checked: "No saturated pixels were detected in any bands used for quantification."

This single sentence preempts one of the most common reviewer objections.

Normalization: What You Divided By and Why

State your normalization strategy explicitly. "Bands were normalized to loading control" is insufficient. Specify:

1. Which loading control: GAPDH (37 kDa), β-actin (42 kDa), vinculin (124 kDa), histone H3 (17 kDa for nuclear fractions), or total protein.
2. Why that control is appropriate: Ideally, cite evidence or your own data showing the control doesn't change with your experimental treatment. If you treated cells with a hypoxia mimetic and normalized to GAPDH — a glycolytic enzyme — a reviewer will (rightly) flag this.
3. If using total protein normalization (TPN): Specify the method — Ponceau S, stain-free (Bio-Rad), REVERT (LI-COR), or Coomassie. State whether you quantified the entire lane or a cropped region. Aldridge et al. (2008) showed TPN can outperform single-protein housekeeping genes in terms of coefficient of variation (CV ~5–10% for TPN vs. ~15–25% for single housekeeping genes), but TPN has its own pitfalls with heavily modified samples or membrane stripping.

A model sentence: "Target band intensity was normalized to total protein per lane, measured by Ponceau S staining of the membrane prior to blocking. Total protein signal was quantified across the full molecular weight range of each lane."

Replicates and Statistics: The Part That Trips People Up

Reviewers want three numbers: how many biological replicates (independent experiments from separate samples or passages), how many technical replicates if applicable, and what statistical test you ran.

Biological replicates: n = 3 independent experiments is the accepted minimum for most journals. Each "n" should be a separate biological preparation — different cell passages, different animals, different patient samples. Running the same lysate on three gels is three technical replicates, not three biological replicates. This distinction matters enormously and reviewers know the difference.

How to report the data: Show individual data points, not just bar graphs. A bar graph with SEM error bars from n = 3 hides the actual spread. Dot plots, or bar graphs with individual points overlaid, are now standard in most high-impact journals. State explicitly: "Data are presented as mean ± SEM of n = [X] independent experiments, with individual data points shown."

Statistical tests: For two-group comparisons (e.g., control vs. treated), an unpaired two-tailed t-test is standard. For multiple groups, use one-way ANOVA with an appropriate post-hoc test (Tukey's for all pairwise, Dunnett's for comparison to control). State the software (GraphPad Prism, R, etc.) and the significance threshold (typically α = 0.05).

One nuance that catches people: normalized western blot data are ratios (target/control), and ratios are not normally distributed — they're log-normal. If you're comparing fold changes across conditions, log-transforming before running your t-test or ANOVA is statistically more appropriate (Kroon et al., 2022). In practice, with n = 3–5 and modest fold changes, it rarely changes the conclusion, but stating "statistical analysis was performed on log₂-transformed fold-change values" is technically more defensible and increasingly expected.

Background Subtraction: A Small Detail Reviewers Notice

Mention your background subtraction method. The options are:

• Local background: A region adjacent to each band, subtracted individually. Most common, but can introduce noise if the background is uneven.
• Rolling ball (ImageJ default): Fits a curve to the background. Works well for gradients.
• Lane-edge or global: Uses the top or bottom of the lane as background. Simpler but assumes uniform background.

One sentence is enough: "Background was subtracted using local background regions adjacent to each band" or "Rolling-ball background subtraction (radius = 50 px) was applied in Fiji."

A Template Methods Paragraph

Here's a complete example you can adapt:

Western blots were detected using ECL (SuperSignal West Pico Plus, Thermo Fisher) and imaged on a ChemiDoc MP (Bio-Rad) with 16-bit acquisition. Exposure times were selected to avoid pixel saturation, confirmed by the overexposure indicator in Image Lab 6.1. Band intensities were measured as integrated density using rectangular ROIs in [software]. Local background was subtracted adjacent to each band. Target protein signal was normalized to vinculin on the same membrane. Vinculin expression did not vary significantly across treatment conditions (one-way ANOVA, p = 0.74). Data are expressed as fold change relative to the vehicle control and presented as mean ± SEM from n = 4 independent experiments. Statistical significance was assessed by unpaired two-tailed Student's t-test (two groups) or one-way ANOVA with Dunnett's post-hoc test (multiple groups), with α = 0.05. Analysis was performed in GraphPad Prism 10.

That's roughly 120 words. It answers every question a reviewer is likely to ask. Copy it, swap in your actual reagents and numbers, and you'll almost certainly avoid a revision request on this point.

What Reviewers Are Really Looking For

In my experience, reviewers don't demand perfection — they demand transparency. They want to know whether your quantification is trustworthy enough to support the claims in your figure legend. The most common reasons quantification gets flagged:

1. No mention of replicates or statistics on quantified blot data.
2. Saturated bands visible in the figures (especially housekeeping controls at high loads — GAPDH saturates easily above ~4 µg total protein with ECL).
3. Bar graphs without individual data points from n = 3, where one outlier could flip the result.
4. "ImageJ" listed without any detail on ROI selection, background subtraction, or normalization.
5. Quantified data from stripped-and-reprobed blots without acknowledging incomplete stripping as a limitation.

Address these proactively and your methods section handles itself. Miss them, and you're looking at a round of revision that costs you a month you didn't need to lose.

References

• Aldridge GM, Podrebarac DM, Greenough WT, Bhatt IH. The use of total protein stains as loading controls: an alternative to high-abundance single-protein controls in semi-quantitative immunoblotting. J Neurosci Methods. 2008;172(2):250–254.
• Butler TA, Paul JW, Chan EC, Smith R, Tolosa JM. Misleading westerns: common quantification mistakes in western blot densitometry and proposed corrective measures. BioMed Res Int. 2019;2019:5214821.
• Gassmann M, Grenacher B, Rohde B, Vogel J. Quantifying western blots: pitfalls of densitometry. Electrophoresis. 2009;30(11):1845–1855.
• Janes KA. An analysis of critical factors for quantitative immunoblotting. Sci Signal. 2015;8(371):rs2.
• Kroon J, et al. Log-transformation and its implications for data analysis. Nat Methods. 2022;19(9):1009–1010.
• Taylor SC, Posch A. The design of a quantitative western blot experiment. BioMed Res Int. 2014;2014:361590.