Clustered Expression Heatmap — Hierarchical Clustering
Cluster and visualize a gene x sample expression matrix with row/column dendrograms.
🌐 Runs on the SeqBench API — also callable via REST & MCP, and in bulk from the batch tools
Paste an expression matrix — genes as rows, samples as columns — and see the classic clustered heatmap: hierarchical clustering reorders both rows and columns so similar genes and similar samples sit next to each other, with dendrograms showing the merge structure. Choose correlation (the standard expression-heatmap default) or Euclidean distance, and average (UPGMA), complete or single linkage; row z-scoring (on by default) puts every gene on a comparable scale so color reflects relative, not absolute, expression, using a blue-white-red diverging scale.
0 gene(s) x 0 sample(s) parsed
Paste an expression matrix (or load the example) to cluster.
Rows are genes, columns are samples. Hierarchical clustering reorders both axes so similar rows/columns sit next to each other, and the optional row z-score puts every gene on a comparable scale so color reflects relative (not absolute) expression.
How to use the Clustered Expression Heatmap tool
- 1Paste a tab-delimited matrix — first row is sample names, first column is gene names — or load the example.
- 2Choose whether to cluster rows/columns, the distance metric and linkage method, and whether to row z-score.
- 3Hover any cell for its gene, sample, raw value and z-score, then export the figure as SVG/PNG.
Frequently asked questions
How does the clustering work?
From-scratch agglomerative (hierarchical) clustering: a full pairwise distance matrix (correlation distance = 1 − Pearson r, or Euclidean), then repeated nearest-pair merging using the standard Lance-Williams update for your chosen linkage (average/UPGMA, complete or single). Leaf order follows the conventional no-crossing left/right dendrogram traversal, so visually adjacent rows/columns are the most similar.
What does row z-score mean, and why is it on by default?
Each gene's values are centered and scaled to its own mean and standard deviation across samples, the conventional 'relative expression' heatmap normalization (as in tools like pheatmap). Without it, genes with a much larger absolute scale would dominate the color range and hide the pattern in lower-expression genes.
How large a matrix can I paste?
Up to 500 genes by 100 samples. Beyond that the clustering and the cell-grid figure get computationally and visually unwieldy — subset your data (e.g. the top variable genes) first.
Is my data stored?
Your pasted matrix is sent to the SeqBench API to compute the clustering and z-scores; it is not stored. The heatmap itself, including hover, renders in your browser, and you can export it as a vector SVG or PNG.