How to Read a Volcano Plot: Fold Change and Significance

A volcano plot is a compact way to scan results from a comparison, often in gene-expression or other high-throughput experiments. Each dot represents a measured feature, such as a gene or protein. Dots farther left or right indicate larger changes between groups; dots higher up generally have stronger statistical evidence against the tested null hypothesis.

Use the plot to identify candidates for closer review, not as proof that a feature matters biologically or that a result is reliable. Before interpreting it, find out which groups were compared, how fold change was calculated, what the vertical axis represents, and whether the analysis adjusted for testing many features.

What You’ll Need

The plot, including its axis labels, legend, and any threshold lines.
The study or analysis notes: group definitions, reference group, sample size, and methods used to calculate fold change and p-values.
If available, the underlying results table with feature names, effect sizes, raw p-values, adjusted p-values, and confidence intervals.

Step-by-Step

Confirm the comparison. Check what the two groups are and which one is the reference or denominator. “Treatment versus control,” for example, can produce the opposite fold-change sign from “control versus treatment.” Also check whether the comparison is adjusted for variables such as batch or age. The plot cannot tell you those details by itself.
Identify each dot. In most plots, one dot is one measured feature. A dot may represent a gene, protein, or another variable, depending on the experiment. Look for labels on selected points or consult the results table; colors and locations alone do not identify a feature.
Read the horizontal axis. It is commonly log2 fold change, though some plots use another effect-size measure. With log2 fold change, zero means no estimated change. Positive values indicate higher measurements in the numerator group; negative values indicate lower measurements in that group. A value of +1 corresponds to about twice the level, and −1 to about half, under that comparison and measurement model. Confirm the axis label before using this interpretation.
Read the vertical axis. It is commonly −log10(p-value). Higher points then have smaller p-values and stronger statistical evidence against the null hypothesis. For example, −log10(p) = 2 corresponds to p = 0.01. This axis is transformed, so equal vertical steps represent multiplicative, not equal absolute, changes in p-value. Some plots use adjusted p-values instead; check the label or legend.
Use position to summarize direction and evidence. Points on the right show positive estimated changes, and points on the left show negative changes; points higher up have stronger evidence if the vertical axis is a negative log p-value. The upper-left and upper-right areas often attract attention because they combine a larger estimated change with stronger evidence. Neither location alone establishes practical importance or reliability.
Check the cutoffs. Horizontal and vertical lines may mark a chosen fold-change threshold and a p-value or adjusted-p-value threshold. A point beyond a line meets that particular cutoff, not a universal definition of significance. Find out whether the rule uses raw p-values or false-discovery-rate-adjusted values, and whether the fold-change cutoff was specified before analysis.
Review multiple-testing adjustment. Testing many features increases the chance of false positives. In large-scale analyses, researchers commonly control the false discovery rate (FDR) and report adjusted p-values, often called q-values. Do not assume a point is an FDR-significant finding just because it is high: verify which value the plot displays and the threshold used.
Look at the full pattern. Notice whether points cluster around zero, whether changes are mostly in one direction, and whether a few extreme points dominate the view. Compare the number of points passing each cutoff on both sides. A lopsided pattern can be real, but it can also reflect the design, normalization, filtering, or plotting choices; investigate rather than infer a cause from the shape alone.
Inspect candidates in the results table. For a point of interest, verify its identifier, estimated change, raw and adjusted p-values, and, if available, confidence interval. Check the underlying measurements and whether the result is consistent across samples. A tiny p-value can accompany a very small effect when measurements are precise, while a large estimated effect can be uncertain when data are sparse or variable.
Validate before drawing a conclusion. Treat the plot as a screening view. Consider study design, data quality, biological or practical context, and independent replication or follow-up evidence. State conclusions with the comparison and cutoff in mind, and distinguish “statistically supported in this analysis” from “important” or “causal.”

Common Mistakes

Reversing the direction: A positive value does not always mean “increased by treatment.” It depends on which group is in the numerator and how the contrast was defined.
Calling every high point significant: Height may reflect a raw p-value, adjusted p-value, or another statistic. Read the axis label and analysis notes.
Treating p-value as effect size: The vertical axis reflects evidence under a statistical model, not how large or useful the change is. Use the horizontal effect-size axis for magnitude.
Ignoring multiple testing: A raw p-value threshold can produce many false positives when thousands of features are tested. Check adjusted results and the stated FDR method.
Assuming a cutoff is universal: Thresholds depend on the question and analysis plan. A color change or guide line is a display choice unless its meaning is documented.
Reading the plot as causal proof: A difference between groups does not show that one feature caused another or that the finding will generalize to other samples.

Tips

Read axis labels, legend, and caption before judging the shape; volcano plots do not all use the same scales.
Use the data table to search by feature name and confirm values rather than estimating exact numbers from a dot’s position.
Keep statistical evidence and practical relevance separate. Consider effect size, uncertainty, and context together.
When comparing plots, check that they use the same contrast direction, transformations, filters, and significance definitions.
Report the threshold and whether it applies to raw or adjusted p-values when describing a set of highlighted features.

Troubleshooting

The plot has no axis labels: Find the figure caption, analysis code, or results table before interpreting positions. Without knowing the scales and contrast, direction and significance may be misread.
No points appear above the threshold: Check whether the plot uses adjusted p-values, whether the cutoff is stringent, and whether the study has enough precision to detect changes. Do not loosen a cutoff just to create findings; report the result as shown.
Points look significant but are not in the table’s hit list: The graph may show raw p-values while the table ranks adjusted p-values, or their cutoffs may differ. Compare the exact values and definitions.
Extreme dots dominate the scale: Confirm they are valid measurements and not data or labeling errors. Use the full table and appropriate quality checks; do not discard points solely because they look unusual.
The left and right sides seem reversed: Check the contrast order and numerator group in the analysis output. If necessary, relabel the comparison clearly before reporting conclusions.

Frequently Asked Questions

What does a volcano plot show?

It displays estimated changes for many features alongside a measure of statistical evidence. Each point usually represents one feature in a group comparison.

What does a point in the upper-right corner mean?

Usually, it indicates a positive estimated change and a small p-value, if the axes are log2 fold change and −log10(p-value). Check the comparison direction and axis definitions first.

Does a high point mean a large change?

No. Height usually indicates statistical evidence, while horizontal distance from zero indicates effect size. A high point can have a small estimated change.

What does a point near zero mean?

It indicates an estimated change near zero on the horizontal scale. It may still be high if the estimate is precise, but the result should be interpreted using its uncertainty and analysis context.

Should I use raw or adjusted p-values?

For analyses testing many features, adjusted p-values are generally important because they account for multiple comparisons. Follow the analysis plan and report clearly which values and cutoff you used.

Can a volcano plot prove a gene or feature is important?

No. It helps prioritize results for review, but importance requires context, sound study design, and often follow-up or independent validation.

Further reading: National Center for Biotechnology Information. Details can change, so check the official page for anything that matters to you. Spotted a mistake? Tell us. How we write an answer.