Volcano Plot Tutorial: Step-by-Step in R, Python, SPSS and StatFAS
Updated November 7, 2026 · StatFAS Tutorials · Beginner to intermediate
This tutorial shows how to create a volcano plot and run the statistics behind the volcano plot in four ways: SPSS, R, Python and the free Volcano Plot Maker on statfas.com. Every script is complete and can be copied into the software. For the concepts behind the volcano plot, read the guide Volcano Plot Explained: Meaning, Uses and Interpretation (R, Python, SPSS).
Step 1: Prepare the data
Genes in rows, samples in columns. Name sample columns by group plus a number (Control_1, Control_2, Treated_1, …). Values should be normalised and on the log2 scale (raw counts are log2(x + 1) transformed automatically on StatFAS when large). Alternatively, upload a results table with the columns Gene, log2FC and pvalue (and optionally padj).
Example layout:
Gene,Control_1,Control_2,Control_3,Treated_1,Treated_2,Treated_3
Gene0001,7.21,7.35,7.10,9.02,8.88,9.15
Gene0002,6.54,6.61,6.48,6.50,6.59,6.66Step 2: SPSS
SPSS has no volcano plot procedure, but the ingredients are available: gene-wise t tests, a Benjamini–Hochberg correction computed with syntax, and a scatter plot built in the Chart Builder.
* Data: one row per sample, one column per gene (Gene0001 ... GeneN), plus GROUP (1 = Control, 2 = Treated).
T-TEST GROUPS=GROUP(1 2)
/VARIABLES=Gene0001 TO Gene0100
/CRITERIA=CI(.95).
* Copy the Welch p values and mean differences into a new dataset (File > Export or OMS),
* giving one row per gene with the variables log2FC and p.
* Benjamini-Hochberg FDR: sort p from largest to smallest, then run:
SORT CASES BY p (D).
COMPUTE m = 100.
COMPUTE rk = m - $CASENUM + 1.
COMPUTE q = MIN(1, p * m / rk).
IF ($CASENUM > 1 AND LAG(q) < q) q = LAG(q).
EXECUTE.
COMPUTE negLog10Q = -LG10(q).
COMPUTE sig = (q < .05 AND ABS(log2FC) >= 1).
EXECUTE.
GRAPH /SCATTERPLOT(BIVAR)=log2FC WITH negLog10Q BY sig.Step 3: R
The code below runs Welch t tests per gene, adjusts the p values and draws the volcano plot with ggplot2. For raw RNA-seq counts, use DESeq2 or edgeR first.
library(ggplot2)
expr <- read.csv("expression.csv", row.names = 1) # genes x samples, log2 scale
grp <- factor(rep(c("Control", "Treated"), each = 6))
res <- as.data.frame(t(apply(expr, 1, function(x) {
tt <- t.test(x[grp == "Treated"], x[grp == "Control"]) # Welch t test per gene
c(log2FC = unname(tt$estimate[1] - tt$estimate[2]), p = tt$p.value)
})))
res$q <- p.adjust(res$p, method = "BH") # Benjamini-Hochberg FDR
res$status <- ifelse(res$q < 0.05 & abs(res$log2FC) >= 1,
ifelse(res$log2FC > 0, "Up", "Down"), "NS")
ggplot(res, aes(log2FC, -log10(q), colour = status)) +
geom_point(size = 1.3) +
scale_colour_manual(values = c(Up = "#d62728", Down = "#1f77b4", NS = "grey60")) +
geom_vline(xintercept = c(-1, 1), linetype = "dashed") +
geom_hline(yintercept = -log10(0.05), linetype = "dashed") +
labs(x = "log2 fold change (Treated vs Control)", y = "-log10(FDR)") +
theme_classic()
ggsave("volcano.png", width = 6, height = 5, dpi = 300)
# RNA-seq raw counts: use DESeq2, then plot with EnhancedVolcano
# library(DESeq2); library(EnhancedVolcano)
# dds <- DESeq(DESeqDataSetFromMatrix(counts, coldata, design = ~ condition))
# r <- results(dds); EnhancedVolcano(r, lab = rownames(r), x = "log2FoldChange", y = "padj")Step 4: Python
The code below uses SciPy for gene-wise Welch t tests, statsmodels for Benjamini–Hochberg correction and Matplotlib for the volcano plot.
import numpy as np, pandas as pd
import matplotlib.pyplot as plt
from scipy import stats
from statsmodels.stats.multitest import multipletests
df = pd.read_csv("expression.csv", index_col=0) # genes x samples, log2 scale
ctrl, trt = df.filter(like="Control"), df.filter(like="Treated")
t, p = stats.ttest_ind(trt, ctrl, axis=1, equal_var=False) # Welch t test per gene
res = pd.DataFrame({"log2FC": trt.mean(axis=1) - ctrl.mean(axis=1), "p": p}, index=df.index)
res["q"] = multipletests(res["p"], method="fdr_bh")[1] # Benjamini-Hochberg FDR
res["sig"] = (res["q"] < 0.05) & (res["log2FC"].abs() >= 1)
plt.figure(figsize=(6, 5))
plt.scatter(res["log2FC"], -np.log10(res["q"]), s=8, c=np.where(res["sig"], "#d62728", "#9a9a9a"))
for x in (-1, 1): plt.axvline(x, ls="--", c="k", lw=0.8)
plt.axhline(-np.log10(0.05), ls="--", c="k", lw=0.8)
plt.xlabel("log2 fold change (Treated vs Control)"); plt.ylabel("-log10(FDR)")
plt.tight_layout(); plt.savefig("volcano.png", dpi=300)
print(res.sort_values("p").head(10))Step 5: The easy way on statfas.com
The Volcano Plot Maker needs no installation and no code. Follow these steps:
- Open the tool. Go to the StatFAS Volcano Plot Maker.
- Add data. Click Load example data to see the layout, then paste or upload an expression matrix (genes by samples) or a results table with log2FC and p values.
- Set thresholds. Choose the |log2FC| threshold, the FDR or p cutoff, the number of labelled genes and the colours.
- Read the results. The tool runs Welch t tests with Benjamini–Hochberg FDR and draws the volcano plot, then writes the APA 7 results text and a top-genes table.
- Download. Download the PNG or SVG figure, or a Word file with the figure, the tables and the interpretation.
Upload or paste data and get the volcano plot, the statistics and an APA 7 write-up in seconds.
Open the free Volcano Plot Maker
Step 6: Interpret and report in APA 7
Check these points before the volcano plot goes into a manuscript:
- Read the sign of the log2 fold change first: positive values mean higher expression in the comparison group, negative values mean lower expression. State the direction in the figure legend.
- Use two thresholds together, for example |log2FC| ≥ 1 (a two-fold change) and FDR < .05. A feature must pass both to count as differentially expressed.
- Plot the FDR (adjusted p value) rather than the raw p value when thousands of genes are tested; raw p values overstate the number of discoveries.
- Check symmetry. A strongly one-sided volcano plot can signal a real biological shift, but also a normalisation problem or a batch effect.
Example APA 7 results paragraph:
Avoid these mistakes with the volcano plot:
- Using raw p values without multiple-testing correction.
- Calculating fold changes on raw counts instead of normalised, log-transformed values.
- Choosing thresholds after looking at the volcano plot to obtain a desired number of hits.
- Labelling only favourite genes instead of the top genes by statistical rank.
Frequently asked questions
What do the axes of a volcano plot mean?
The x axis is the log2 fold change between two groups, and the y axis is the negative log10 of the p value or FDR. Points far from zero horizontally show large changes; points high up show strong statistical evidence.
Which cut-offs should a volcano plot use?
Common choices are |log2 fold change| ≥ 1 and FDR < .05. Choose thresholds before looking at the data and report them in the methods and the figure legend.
Should a volcano plot use p values or FDR?
Use the FDR (for example Benjamini–Hochberg) when many genes are tested, because the FDR controls the expected proportion of false discoveries among the genes called significant.
How many replicates does a volcano plot need?
Three per group is a bare minimum for RNA-seq; five or more per group gives far more stable results. Power depends on effect size and variability.
Related guides
How to cite this page: StatFAS. (2026, November 7). Volcano Plot Tutorial: Step-by-Step in R, Python, SPSS and StatFAS. https://statfas.com/tutorials/volcano-plot.html