A simulated single-cell RNA sequencing heatmap showing 200 differentially expressed genes across 150 cells from a PBMC (peripheral blood mononuclear cell) dataset, clustered into 6 cell types.
Values are z-scores (row-normalized) derived from log2-transformed TPM expression values, following standard RNA-seq analysis conventions. The diverging blue-white-red color scale represents downregulation (blue, z < 0) through neutral (white, z = 0) to upregulation (red, z > 0), clamped at +/- 3 SD.
In a real analysis: FASTQ reads are aligned (HISAT2/STAR), quantified (featureCounts), normalized (TPM/CPM), tested for differential expression (DESeq2), then the top DEGs are z-score transformed and hierarchically clustered for visualization.
Tools like Vitessce and cellxgene display gene labels as all-or-nothing: either all 20,000 names are visible (illegible) or none. Pretext enables a middle ground: layout() computes exactly which genes have enough vertical space for their name at the current row height. 200 layout calls in under 0.05ms.
Hover any gene row to expand it. The gene name, function, and expression statistics flow into the expanding space, with Pretext determining how much detail text fits in real time.
Gene markers are real immune cell surface proteins and transcription factors (CD3D, CD19, LYZ, HBB, GNLY, etc.) with biologically plausible cluster-specific expression patterns. Cell types: CD4+ T, CD8+ T, B cell, NK, Monocyte, Dendritic cell.
Built with Pretext by Cheng Lou. Analysis pipeline references: OLV Tools.