MA Plot
A scatter plot that compares log-ratio intensity (M) against mean average intensity (A) across two conditions — the workhorse of genomic normalization quality control.
// 01 — The chart
What it looks like
An MA plot comparing treated vs. control samples. Red dots show up-regulated genes; green dots show down-regulated genes. The majority of points cluster around M = 0 (no change).
// 02 — Definition
What is an MA plot?
An MA plot (also known as an RI plot or Bland–Altman plot in statistics) is a scatter plot designed for comparing measurements from two experimental conditions. The name comes from its axes: M for the log-ratio (minus) and A for the average.
In genomics, each dot represents a single gene or transcript. The horizontal axis (A) shows the average expression level across both conditions — how abundant the gene is overall. The vertical axis (M) shows the log₂ fold change between the two conditions — how much the gene changed.
Well-normalized data should produce a cloud of points centered on M = 0, meaning most genes are unchanged. Points far above or below this line are the candidates for differential expression — the genes that truly changed between conditions.
Origin: The MA plot was popularized in bioinformatics by Yang et al. (2002) for visualizing microarray data. Its statistical ancestor, the Bland–Altman plot (1986), was developed to compare two measurement methods in clinical chemistry.
// 03 — Anatomy
Parts of an MA plot
// 04 — Usage
When to use it — and when not to
- Comparing expression levels between two conditions (treated vs. control)
- Assessing whether normalization removed intensity-dependent bias
- Identifying differentially expressed genes or proteins
- Checking for systematic trends that depend on expression level
- Visualizing microarray, RNA-seq, or proteomics fold-change data
- You want to show statistical significance — use a volcano plot instead
- Your data involves more than two conditions — use a heatmap
- You need to display time-series gene expression changes
- Audience is non-technical — the log-ratio axis is hard to interpret
- You have only a handful of features — a bar chart is more intuitive
// 05 — Reading guide
How to read an MA plot
Follow these steps to interpret any MA plot you encounter.
Identify the axes
The horizontal A-axis shows mean average expression (how abundant a gene is). The vertical M-axis shows the log₂ fold change between conditions. M = 0 means no change; M = 1 means a 2-fold increase.
Check the central cloud
Most points should cluster symmetrically around M = 0. If the cloud drifts up or down, it signals a normalization issue or global shift in expression between samples.
Look for intensity-dependent bias
If the cloud fans out or curves at low A values (left side), low-expression genes have more variability. This ‘funnel shape’ is expected but should not show systematic skew.
Find differentially expressed features
Points far above or below the fold-change thresholds are candidates. Up-regulated genes sit above the upper dashed line; down-regulated genes sit below the lower line.
Evaluate normalization quality
A well-normalized dataset will have a loess smoothing curve that lies flat along M = 0 across the full range of A values. A curved or tilted loess line means normalization needs adjustment.
// 06 — Pitfalls
Common mistakes
Ignoring the funnel shape at low A
Low-expression genes naturally have higher variance. Calling them differentially expressed without statistical testing leads to false positives.
Skipping the loess smoothing curve
Without a loess fit overlaid on the data, you cannot assess whether normalization was successful. Always add one.
Using raw fold change instead of log-ratio
Raw ratios compress down-regulated genes (between 0 and 1) while stretching up-regulated ones. The log transformation ensures symmetry.
Treating the MA plot as a significance test
The MA plot shows magnitude of change, not statistical significance. Always pair it with a volcano plot or adjusted p-values before calling a gene significant.
Overplotting without transparency
With thousands of genes, points overlap heavily. Use semi-transparent dots or 2D density coloring to reveal the true distribution.
// 07 — In the wild
Real-world examples
Cancer genomics
Comparing tumor vs. normal tissue RNA-seq samples to identify oncogenes and tumor suppressors with strong fold changes.
Drug response studies
Pharmaceutical teams use MA plots to compare gene expression before and after drug treatment, assessing dose-dependent transcriptomic changes.
Microarray quality control
Core genomics facilities plot MA plots for every pair of biological replicates to detect batch effects, dye bias, or failed hybridizations.
Single-cell RNA-seq
Comparing cell clusters to identify marker genes that distinguish one cell type from another in complex tissues.
// 08 — At a glance
Quick reference
| Also known as | RI plot, Bland–Altman plot (statistics), ratio–intensity plot |
| Primary use | Assess normalization quality and identify differentially expressed features |
| Data type | Two-condition quantitative measurements (expression values, intensities) |
| X-axis (A) | Mean average expression = ½(log₂ R + log₂ G) |
| Y-axis (M) | Log-ratio = log₂(R/G) = log₂ R – log₂ G |
| Tools | R (limma, DESeq2, edgeR), Python (pyDESeq2, bioinfokit), Bioconductor |
| Difficulty | Advanced — requires understanding of log transformations |
// 09 — Variations
Variations & relatives
Smear plot
An MA plot variant in edgeR that highlights low-count genes differently and uses average log-CPM on the A-axis.
MD plot
Mean-difference plot used in limma. Identical to MA but with different naming convention (M = log-fold change, D = difference).
MA plot with loess
Adds a loess smoothing curve to visualize systematic bias across expression levels. Essential for normalization assessment.
Density MA plot
Colors points by local density (2D kernel density estimate) to handle overplotting in large datasets with millions of features.
// 10 — FAQs
Frequently asked questions
What is a ma plot?+
An MA plot (also known as an RI plot or Bland–Altman plot in statistics) is a scatter plot designed for comparing measurements from two experimental conditions. The name comes from its axes: M for the log-ratio (minus) and A for the average.
When should you use a ma plot?+
Use a ma plot when comparing expression levels between two conditions (treated vs. control). It also works well when assessing whether normalization removed intensity-dependent bias, and when identifying differentially expressed genes or proteins.
When should you avoid a ma plot?+
Avoid a ma plot when you want to show statistical significance — use a volcano plot instead. It is also a poor fit when your data involves more than two conditions — use a heatmap, or when you need to display time-series gene expression changes.
Is a ma plot suitable for dashboards?+
Yes — a ma plot can work well in dashboards as long as the panel is large enough for readers to perceive the encoded values, has a clear title, and includes the legend or axis labels needed to interpret it.
What category of chart is a ma plot?+
MA Plot belongs to the Scientific family of charts. Charts in that family are designed to answer the same kind of question, so they often work as alternatives when one doesn't quite fit your data.
How do you read a ma plot?+
Start with the axis labels and legend, then look at the overall shape before zooming into individual marks. Compare prominent features against the rest of the data, and verify any conclusion against the underlying numbers when precision matters.