Help & Documentation

STARK (Single-cell Transcription stARt site Knowledgebase) is a comprehensive single-cell TSS (Transcription Start Site) database browser. It integrates single-cell RNA-seq data from multiple cancer types to provide detailed insights into TSS usage, gene expression, and cell-type specificity.

The database currently contains 428 samples across 14 cell types, with a total of over 2.2 million cells. Each sample has been processed through a standardized pipeline including UMAP embedding, cell-type annotation, and TSS quantification.

Use the navigation bar above to explore different aspects of the data. Each section is described below.

The Samples Browser page provides a searchable, filterable table of all 428 samples in the database.

• Search by keyword across sample titles.
• Filter by SRR, Study Accession, Source, Cancer type, and Tissue.
• Sort any column by clicking the column header.
• Select rows with checkboxes and Export to CSV.
• Click an SRR to view the Sample Detail page.

The TSS Browser page lists all transcription start sites (~41K records) identified across the dataset.

• Search by TSS ID or Gene Symbol with autocomplete suggestions.
• Filter by TSS Type (Canonical / Novel) and Chromosome.
• Sort any column by clicking the column header.
• Export selected or all records to CSV.
• Click a TSS ID to view the TSS Detail page.

The Gene Browser page provides an overview of genes and their associated TSS variants.

• Search by Gene Symbol with autocomplete suggestions.
• Filter by Chromosome and Strand.
• Sort any column by clicking the column header.
• Export selected or all records to CSV.
• Click a Gene Symbol to view the Gene Detail page.

The Map section provides aggregated statistics and visualizations grouped by Tissue or Cancer type.

• Tissue Map — browse samples by tissue of origin.
• Cancer Map — browse samples by cancer type.
• Each detail page shows summary cards, an integrated UMAP plot, and a sample list.
• The UMAP can be colored by Cell Type or Sample ID.

The Download page allows you to browse samples and download their associated TSS annotated .h5ad files (compact format).

• Search and filter samples by various criteria.
• File availability is automatically checked.
• Click the TSS button to download the corresponding .h5ad file.

The Sample Detail page (accessible from the Samples Browser) provides an in-depth view of a single sample.

• Basic Information — metadata including tissue, cancer type, and cell count.
• UMAP — interactive scatter plot of cell embeddings colored by cell type.
• Cell Type Composition — donut chart showing cell type proportions.
• TSS QC — violin and box plots for TSS quality metrics.
• RNA/TSS Evidence — bubble plot showing gene expression and TSS usage across cell types.
• Download — direct links to TSS .h5ad files.

The TSS Detail page (accessible from the TSS Browser) provides comprehensive information about a single transcription start site.

• Basic Information — genomic coordinates, strand, and gene annotation.
• Expression UMAP — overlay of TSS-expressing cells on the sample background UMAP.
• Cell Type Composition — pie chart of cell types expressing this TSS.
• TSS Expression by Cell Type — violin plots showing expression distribution per cell type.
• TSS-Gene Correlation — scatter plot with regression line showing RNA vs TSS UMI correlation.
• Sample Expression Details — table of expression metrics per sample.
• IGV Browser — genome browser view of the TSS locus.

The Gene Detail page (accessible from the Gene Browser) provides information about a gene and its associated TSS variants.

• Basic Information — gene symbol, ID, and genomic coordinates.
• TSS Variants — table of all TSS variants for this gene, linked to their TSS Detail pages.
• TSS Promoter Usage — stacked bar chart showing TSS usage proportions across cell types.