Here we provide a sample by sample correlation analysis heatmap. The correlation analysis heatmap highlights the differences and similarity between samples based on how strongly or weakly they correlate with each other across all gene expression values. Generally we would expect replicates to have similar expression patterns across all genes, and so we would expect replicates to have high correlation values. Similarly we might expect very different sample groups to have very different expression patterns and so to have low correlation values. This is reflected in the heatmap, which shows all possible combinations of correlations (Spearman Correlation Coefficient) between two samples. The diagonal line indicates that when comparing a sample to itself we should always get a correlation of 1. 
<br />
<br />
The left and right plots are identical, however in the right hand plot the samples have been hierarchically clustered, so that more similar samples are grouped together on the x and y axis. We would expect replicate samples to group after clustering and cohorts to separate. Hierarchical clustering can therefore provide an extra level of evidence for sample similarity / dissimilarity and can help to clarify results visually.

