Tool for the analysis and visualization of sample correlations based on the output of multiBamCoverage or multiBigwigSummary. Pearson or Spearman methods are available to compute correlation coefficients. Results can be saved into a heatmap image or as multiple scatter plots depicting the pairwise correlations. Optionally, the values can be saved as tables, too.

detailed help:

plotCorrelation -h

usage: plotCorrelation [-h] --corData FILE --plotFile FILE --corMethod
                       {spearman,pearson} --whatToPlot {heatmap,scatterplot}
                       [--labels sample1 sample2 [sample1 sample2 ...]]
                       [--plotTitle PLOTTITLE] [--plotFileFormat FILETYPE]
                       [--removeOutliers] [--version]
                       [--outFileCorMatrix FILE] [--zMin ZMIN] [--zMax ZMAX]
                       [--colorMap] [--plotNumbers]
Required arguments
--corData, -in Compressed matrix of values generated by multiBigwigSummary or multiBamCoverage
--plotFile, -o File to save the heatmap to. The file extension determines the format, so heatmap.pdf will save the heatmap in PDF format. The available formats are: .png, .eps, .pdf and .svg.
--corMethod, -c

Correlation method.

Possible choices: spearman, pearson

--whatToPlot, -p

Choose between a heatmap or pairwise scatter plots

Possible choices: heatmap, scatterplot

Optional arguments
 By setting this option, genomic regions that have zero or missing (nan) values in all samples are excluded.
--labels, -l User defined labels instead of default labels from file names. Multiple labels have to be separated by spaces, e.g. –labels sample1 sample2 sample3
--plotTitle=, -T=
 Title of the plot, to be printed on top of the generated image. Leave blank for no title.

Image format type. If given, this option overrides the image format based on the plotFile ending. The available options are: png, eps, pdf and svg.

Possible choices: png, pdf, svg, eps

 If set, bins with very large counts are removed. Bins with abnormally high reads counts artificially increase pearson correlation; that’s why, by default, multiBamCoverage tries to remove outliers using the median absolute deviation (MAD) method applying a threshold of 200 to only consider extremely large deviations from the median. The ENCODE blacklist page ( contains useful information about regions with unusually high counts.
--version show program’s version number and exit
Output optional options
 Save matrix with pairwise correlation values to a tab-separated file.
Heatmap options
--zMin, -min Minimum value for the heatmap intensities. If not specified, the value is set automatically
--zMax, -max Maximum value for the heatmap intensities.If not specified, the value is set automatically
--colorMap=jet Color map to use for the heatmap. Available values can be seen here:
 If set, then the correlation number is plotted on top of the heatmap. This option is only valid when plotting a heatmap.

example usages: plotCorrelation -in results_file –whatToPlot heatmap –corMethod pearson -o heatmap.png


plotCorrelation computes the overall similarity between two or more files based on read coverage (or other scores) within genomic regions, which must be calculated using either multiBamCoverage or multiBigwigSummary.

The result of the correlation computation is a table of correlation coefficients that indicates how “strong” the relationship between two samples is and it will consist of numbers between -1 and 1. (-1 indicates perfect anti-correlation, 1 perfect correlation.)


We offer two different functions for the correlation computation: Pearson or Spearman.

The Pearson method measures the metric differences between samples and is therefore influenced by outliers. The Spearman method is based on rankings. If you imagine a race with 3 participants where the winner and runner-up are very close together while the third person broke her leg and comes in way, way after the first two, then Pearson would be strongly influenced by the fact that the third person had a great distance to the first ones while Spearman would only care about the fact that person 1 came in first, person 2 came in second and person 3 got the third rank, the distances between them are ignored.


Pearson is an appropriate measure for data that follows a normal distribution, while Spearman does not make this assumption and is generally less driven by outliers, but with the caveat of also being less sensitive.

Here’s an example of RNA-seq data from different human cell lines that we had downloaded from


As you can see, both correlation calculations more or less agree on which samples are nearly identical (the replicates, indicated by 1 or 2 at the end of the label). The Spearman correlation, however, seems to be more robust and meets our expectations more closely as the two different cell types (HUVEC and IMR90) are clearly separated.


In the following example, a correlation analysis is performed based on the coverage file computed by multiBamCoverage or multiBigwigSummary for our test ENCODE ChIP-Seq datasets.


Here we make pairwose scatterplots of the average scores per transcript that we calculated using multiBigwigSummary and include the Pearson correlation coefficients for each comparison.

$ deepTools2.0/bin/plotCorrelation \
-in scores_per_transcript.npz \
--corMethod pearson --skipZeros \
--plotTitle "Pearson Correlation of Average Scores Per Transcript" \
--whatToPlot scatterplot \
-o scatterplot_PearsonCorr_bigwigScores.png   \
$ cat
    'H3K27me3'      'H3K4me1'       'H3K4me3'       'HeK9me3'       'input'
    'H3K27me3'      1.0000  -0.1032 -0.1269 -0.0339 -0.0395
    'H3K4me1'       -0.1032 1.0000  0.3985  -0.1863 0.3328
    'H3K4me3'       -0.1269 0.3985  1.0000  -0.0480 0.2822
    'HeK9me3'       -0.0339 -0.1863 -0.0480 1.0000  -0.0353
    'input' -0.0395 0.3328  0.2822  -0.0353 1.0000


In addition to scatterplots, heatmaps can be generated where the pairwise correlation coefficients are depicted by varying color intensities and are clustered using hierarchical clustering.

The example here calculates the Spearman correlation coefficients of read counts. The dendrogram indicates which samples’ read counts are most similar to each other.

$ deepTools2.0/bin/plotCorrelation \
    -in readCounts.npz \
    --corMethod spearman --skipZeros \
    --plotTitle "Spearman Correlation of Read Counts" \
    --whatToPlot heatmap --colorMap RdYlBu --plotNumbers \
    -o heatmap_SpearmanCorr_readCounts.png   \

plotCorrelation in Galaxy

Below is the screenshot showing how to use plotCorrelation with deepTools Galaxy.