MCAT content library / Research & data / Normalized data, percent change, and fold change
Recover the numerator, denominator, and reference behind a normalized result.
About 14 minutes, including the worked examples and checks.
This lesson appears in more than one study path. Follow the links named for your chosen path to stay in that sequence.
From a rate curve to a mechanism · Lesson 4 of 4 · Continue in this path
Read a study and its figures · Lesson 5 of 6 · Continue in this path
On this page
- Concept: a normalized value answers a particular question
- Pattern: normalize the sample, then compare with a reference
- Application: count what changed before dividing
- Rule: choose a denominator that serves the biological question
- Application: translate between fold and percent change
- Check your reasoning
- Continue learning
- Sources
Concept: a normalized value answers a particular question
A treated culture produces three times as much target-protein signal as a control culture. Does each cell contain three times as much target protein? Only if the cultures contain comparable numbers of cells and the measurement can be interpreted that way. A larger sample can produce a larger signal even when its individual cells have not changed.
Normalization expresses a measurement relative to another quantity, such as cell number, total protein, or a reference condition. It can make a comparison more informative by accounting for a specified difference between samples. It also changes the quantity being compared. “Signal per cell,” “signal per microgram of protein,” and “total culture signal” answer different biological questions.
AAMC data reasoning includes evaluating representations and drawing conclusions that the supplied measurements support. For normalized data, that begins by recovering the numerator and denominator behind the reported value.1
Pattern: normalize the sample, then compare with a reference
Let S denote measured target signal and D the chosen denominator. The normalized quantity is N = S/D. To express treatment relative to control, calculate fold change = Ntreatment/Ncontrol. Written out, this is (Streatment/Dtreatment)/(Scontrol/Dcontrol).
Two divisions are involved. The first defines what each sample represents; the second defines the reference comparison. If a graph simply says “relative expression,” read the caption to identify both operations. A control value of 1 may mean that a reference ratio has been set to 1, not that every control cell contains one molecule.
Signal ratios also require a valid measurement scale. In quantitative protein assays, a twofold signal is a useful proxy for a twofold amount only within an appropriate response range, after suitable background treatment. Saturated detection can flatten a real difference. Experimental work on quantitative western blotting demonstrates why the target signal and any loading-control signal need their own validated quantitative ranges.2
Application: count what changed before dividing
The following dataset is illustrative. Each row describes a complete culture. Signals are already background-corrected, measured using the same assay, and within its linear range. Cell counts and total protein amounts refer to the same harvested material.
| Culture | Cells (millions) | Target signal (AU) | Total protein (μg) |
|---|---|---|---|
| Control | 1.0 | 100 | 20 |
| Treatment A | 2.0 | 300 | 40 |
| Treatment B | 1.0 | 200 | 40 |
For Treatment A, total signal is 300/100 = 3 times control. But the cell count is also doubled. Signal per million cells is 300/2 = 150 AU, compared with 100/1 = 100 AU for control. The per-cell fold change is 150/100 = 1.5. This culture has three times the total target signal and 1.5 times the target signal per cell. Both statements can be true.
Normalizing to total protein gives 300/40 = 7.5 AU/μg for Treatment A and 100/20 = 5 AU/μg for control, again a 1.5-fold ratio. The two normalization choices agree here because total protein per cell is the same in these two cultures.
| Culture | Signal per million cells (AU) | Signal per μg protein (AU/μg) | Fold change per cell | Fold change per μg protein |
|---|---|---|---|---|
| Control | 100 | 5 | 1.0 | 1.0 |
| Treatment A | 150 | 7.5 | 1.5 | 1.5 |
| Treatment B | 200 | 5 | 2.0 | 1.0 |
Treatment B shows why the denominator matters. It has the same number of cells as control but twice the total protein. Its target signal per cell doubles, while its target signal per microgram of total protein stays constant. Saying “there is no change in target protein” discards the per-cell result. Saying “target protein is unchanged relative to total protein” identifies the comparison that actually stayed constant.
Rule: choose a denominator that serves the biological question
If the question concerns average target abundance per cell, cell number is directly relevant. If it concerns the target’s abundance relative to the sampled protein pool, total protein addresses that ratio. Neither denominator is universally correct. The question and the sample preparation determine what the normalized quantity means.
A reference protein is sometimes used to account for sample loading. That interpretation assumes its amount is suitable as a stable reference across the compared conditions. A treatment that changes the reference can distort the target/reference ratio. A consistent-looking band is also not enough if its signal is saturated; measured consistency could reflect a detector limitation. These are empirical properties to validate, not guarantees supplied by the term “housekeeping protein.”2
Background correction comes before interpreting ratios. If a shared background of 20 AU is included in readings of 120 and 220 AU, corrected signals are 100 and 200 AU. Their ratio is 2. The uncorrected ratio, 220/120 ≈ 1.83, mixes target signal with the additive background. The correction is justified here because the background was specified and measured; it should not be invented to obtain a preferred result.
Application: translate between fold and percent change
For a positive reference value, percent change = (new − reference)/reference × 100%. A fold change F corresponds to (F − 1) × 100%. Therefore, 1.5 times the reference is a 50% increase; 0.75 times the reference is a 25% decrease. State “1.5 times the reference” or “50% higher” to make the meaning explicit.
The reference changes when the direction of comparison changes. Moving from 100 to 150 is a 50% increase, but returning from 150 to 100 is a decrease of 50/150 ≈ 33.3%. The absolute change is 50 in each direction; the percentage differs because its denominator differs.
A log2 fold-change axis expresses multiplicative changes symmetrically: a doubling has log2(2) = +1; a halving has log2(0.5) = −1; no change has log2(1) = 0. This zero means equal compared values, not absence of the target. The real logarithm requires a positive ratio.3 A zero reference does not yield an ordinary finite fold change, and adding a small constant changes the analysis rather than repairing it invisibly.
Finally, a fold-change graph can hide absolute amounts. Going from 10 to 15 units and from 100 to 150 units both gives 1.5-fold, despite increases of 5 and 50 units. Recover the baseline before comparing absolute changes. Keep the reference visible in your interpretation: what changed, per what, and relative to which condition?
ACTIVE PRACTICE · COMPARE AND TRANSFER
Choose the Denominator Before Calculating Fold Change
Original background-corrected target measurements: control 1.0 million cells, signal 100 AU, total protein 20 µg; treatment A 2.0 million cells, 300 AU, 40 µg; treatment B 1.0 million cells, 200 AU, 40 µg. All signals are within the validated linear assay range.
Examine the original evidence
Results are bounded to the original supplied observations. Instructional checks do not establish assessment calibration or clinical effectiveness.
Check your reasoning
Choose an answer before opening its feedback. Each option explains the reasoning behind that choice.
Check 1
A control culture contains 0.50 million cells and produces 80 AU of target signal. A treated culture contains 1.00 million cells and produces 120 AU. The signals are background-corrected and within a common linear range. What is the treated-to-control fold change in signal per cell?
A. 0.50
Review this choice. The treated culture has twice as many cells, but it also has 1.5 times the total signal. Both numerator and denominator changes must be included.
B. 0.75
Correct. Control signal per million cells is 80/0.50=160 AU. Treated signal per million cells is 120/1.00=120 AU. Their ratio is 120/160=0.75, a 25% decrease per cell.
C. 1.50
Review this choice. 1.50 is 120/80, the ratio of total signals. It ignores the doubled cell count.
D. 3.00
Review this choice. Multiplying by the cell-count ratio moves in the wrong direction. To obtain signal per cell, divide each signal by its own cell count.
The principle to carry forward
A larger total can coexist with a lower value per cell. Define the sample-level quantity before calculating its fold change.
Check 2
Two experiments use the same calibrated units. Experiment 1 reports a control value of 20 and a treated value of 30. Experiment 2 reports 80 and 120. Both figures set their own control to 1. Which comparison is correct?
A. Both experiments show the same absolute increase and the same fold change.
Review this choice. The fold changes agree, but the absolute increases are 10 and 40 units. Setting both controls to 1 hides that difference.
B. Experiment1 has the larger fold change because it begins at a lower baseline.
Review this choice. A lower baseline does not imply a larger ratio when the numerator changes proportionally. Here 30/20 and 120/80 both equal 1.5.
C. Experiment2 has four times the absolute increase, with the same fold change.
Correct. The absolute increases are 30−20=10 and 120−80=40. Their fold changes are both 1.5. Common calibration makes the absolute-unit comparison meaningful.
D. Experiment2 has four times the fold change, with the same absolute increase.
Review this choice. This reverses the comparisons. The absolute increase is four times as large, while both treatment/reference ratios are 1.5.
The principle to carry forward
Equal relative changes do not establish equal absolute changes; comparing absolute units also requires comparable calibration.
Continue learning
From a rate curve to a mechanism
Lesson 4 of 4. You have reached the end of this path.
- Previous in From a rate curve to a mechanism: Experimental Controls: What a Comparison Can Show
- Return to this path’s overview
See every step in From a rate curve to a mechanism
From a rate curve to a mechanism
Connect the existing kinetics lesson to regulation, valid comparisons, and normalized results.
- Enzyme kinetics: read the curve before naming the inhibitor
- How enzymes catalyze reactions and regulate pathways
- Experimental Controls: What a Comparison Can Show
- Normalized data, percent change, and fold change · You are here
Read a study and its figures
Lesson 5 of 6.
- Previous in Read a study and its figures: Reading graphs: axes, scales, and the comparison that matters
- Next in Read a study and its figures: Mean, median, and spread: describing a dataset
- Return to this path’s overview
See every step in Read a study and its figures
Read a study and its figures
Follow a question through its design, graph, normalization and statistical description.
- Hypotheses, predictions, and competing explanations
- Experimental Controls: What a Comparison Can Show
- Observational and experimental studies: what can we conclude?
- Reading graphs: axes, scales, and the comparison that matters
- Normalized data, percent change, and fold change · You are here
- Mean, median, and spread: describing a dataset
Related lessons
- Reading graphs: axes, scales, and the comparison that matters
- Experimental Controls: What a Comparison Can Show
- Mean, median, and spread: describing a dataset
- Choosing and interpreting protein purification methods
Browse research & data · MCAT content library
MCAT preparation and roadmap · Aim High, Score Higher
Sources
1. AAMC: Scientific Inquiry and Reasoning Skill 4
2. Taylor et al.: A Defined Methodology for Reliable Quantification of Western Blot Data