How to use the cohort explorer
Layout
- One row per sample; the genome (hg38) runs left → right. Blue = copy-number loss, red = gain (log2 ratio), hatched = LOH.
- Dots are driver mutations at their genomic position, coloured by variant class, sized by VAF.
- The metadata heatmap on the left shows clinical attributes and an oncoprint-style mutation matrix (group Mutations).
- The strip above the samples is the copy-number landscape of the current group(s): mean gain (red) and loss (blue) per position. It repeats per group — with many groups hide it via the eye menu (tracks → CN landscape) to give the rows more space.
Interact
- Right-click a metadata column header or cell → sort, filter (retain/remove), group by that attribute, group by quartiles/thresholds.
- Right-click the genomic panel → sort samples by the copy number / mutation at that locus (e.g. sort by chr7 loss).
- Scroll to zoom the genome, drag to pan. Type a gene (
GATA1), band (7q36) or region (chr8:127,000,000-128,000,000) into the search box.
- Type a metadata value (e.g.
ML-DS) into the search box to filter samples by it.
- Press E or use the context menu to peek at samples in a close-up; Ctrl+wheel still zooms.
Save & share
- Every action is in the undo history (toolbar). The provenance menu lists the steps.
- Bookmark a state (stored in your browser) or use Share to copy a URL that reproduces it.
- The eye icon toggles tracks and metadata groups; the ⋮ menu exports PNG.
- Import your own sample table via Upload metadata (TSV/CSV, first column = sample id).
Data: synthetic demonstration cohort — see cohort/BUILD.md for how the real export is wired in.