The Cancer cohort menu lists TCGA and CPTAC cohorts. The same cancer in two studies is two separate groups. For example, TCGA-PAAD and CPTAC-3-PAAD are never combined.
Enter a gene symbol, such as TP53, or an Ensembl ID. The menu shows each cohort's patient count. A plot can use fewer patients when expression or follow-up is missing.
Each comparison makes two groups: lower and higher expression.
| Choice | Compares | Left out |
|---|---|---|
| Median | At or below the middle value vs above | No one |
| Mean | At or below the average TPM vs above | No one |
| Percentile | Below vs above a point you pick, such as 75% | No one |
| Lowest/highest quarter | Bottom 25% vs top 25% | The middle |
| Lowest/highest third | Bottom third vs top third | The middle |
| Custom percentile groups | For example, lowest 20% vs highest 30% | The middle |
| Custom TPM threshold | At or below your number vs above | No one |
- TPM measures a gene's RNA level in a sample.
- Equal values stay in the same group. So a "quarter" may not be exactly 25%, and a group can be empty. Check the group sizes before you plot.
- Each outcome uses only patients with follow-up for it. Cutoffs and sizes can differ by outcome.
- Groups that add up to 100% (such as 30%/70%) become one split with no one left out.
Select Create survival plot after each change. A notice shows when your choices differ from the plot. SurvScope never hunts for the cutoff with the smallest p-value. Details are in methods.
A Kaplan–Meier curve starts at 100%. It steps down at each event. A higher curve means more patients are still event-free.
| Panel | Event |
|---|---|
| OS, overall survival | Death from any cause |
| DSS, disease-specific survival | Death from the cancer |
| PFI, progression-free interval | Progression, as defined by TCGA-CDR |
| DFI, disease-free interval | New tumor after being disease-free (TCGA-CDR) |
CPTAC has OS only. Its other panels say Endpoint unavailable. Take care with outcomes marked caution or not recommended. Very small groups may have no result.
| Label | Meaning |
|---|---|
| n | Patients in the group |
| e | Patients with an event. The rest are censored (follow-up ended with no event). So e ≤ n. |
| p | Log-rank p-value for the difference between the curves |
| q | Optional adjusted p across this analysis's outcomes. See q-values. |
| HR | Hazard ratio, higher vs lower expression. Above 1 means more events in the higher group. Not adjusted for stage, age, treatment or other factors. |
| NA | Cannot be estimated, for example with an empty group or no events |
Properties → Survival details can add confidence bands (90%, 95% or 99%), censor marks (+) and number-at-risk tables. All are off by default.
Results are associations, not proof of cause.
- Double-click any text to edit it. Select characters for bold, italic, superscript or subscript. Enter adds a line. Escape finishes.
- Click an object to select it, or pick it from Selected item. Drag to move. Use the corner handle to resize.
- Shift-click or drag a box to select several. Arrange aligns and spaces them. Snap shows guides.
- In Layers, the eye hides an object and the padlock locks it.
- Hover over an icon to see its name and shortcut.
- Analysis is on the left and Properties / Layers on the right. Drag their edges to resize. Reset workspace resets the panels, not the figure.
| Key | Action |
|---|---|
| V / T / H / Z | Selection / Type / Hand / Zoom |
| Space (hold) | Pan |
| Arrow / Shift+Arrow | Move 1 / 10 points |
| Ctrl/Cmd+Z / Ctrl/Cmd+Shift+Z | Undo / redo |
| Ctrl/Cmd+C / V / D | Copy / paste / duplicate annotations |
| Ctrl/Cmd+B / I | Bold / italic while typing |
| Delete | Remove annotations |
With Type, click an empty spot to add a note. Drag with Line or Arrow to draw one.
| Properties section | What it sets |
|---|---|
| Size and appearance | Size in inches, font, text size, colors, line width and style |
| Outcomes and layout | Which panels show, their grid and order (display only) |
| Axes | Time unit, time range, ticks, survival range (display only) |
| Legend and annotations | Group names, legend text, notes, lines, arrows |
Legend edits change the figure only. JSON and citations keep the real counts. Reset selected item restores the automatic text.
Fit, 100% and zoom change the view, not the export size. Reset figure restores the default look and can be undone.
A new gene, cohort or comparison keeps your style. It clears custom labels and notes. Save a project first to keep them.
Use Size and appearance → Font for the whole figure, or Selected font for one text object.
| Menu choice | Font used |
|---|---|
| Arial / Helvetica style | Liberation Sans (default) |
| Times New Roman style | Liberation Serif |
| Courier New style | Liberation Mono |
| Calibri style | Carlito |
| Lato | Lato |
| Source Sans 3 | Source Sans 3 |
| Source Serif 4 | Source Serif 4 |
The "style" fonts are look-alikes, not the original fonts. Exports embed the fonts, so figures look the same on any computer. See font sources.
| Button | Gives you |
|---|---|
| SVG | Vector figure with embedded fonts |
| Vector figure with embedded fonts, at your page size | |
| PNG | Image at 150, 300 or 600 DPI (set in Export and reuse) |
| JSON | Results and data sources, without the layout |
| Save project | Results, curves, labels, notes, style and versions |
| Save style preset | Style only |
6.8 inches at 300 DPI gives a 2040 × 2040 PNG. The PNG limit is 40 million pixels.
Open project opens a project or preset. Files stay on your computer. A project keeps its results, even from an older data release. Choose New analysis to use current data.
A URL saves the gene, cohort and comparison. Share a project file to share the edited figure. Projects contain no patient IDs or expression data.
Choose Cite this analysis. Use Copy references, Download BibTeX or Download RIS. The methods text lists gene, cohort, grouping, counts, exclusions and versions. If no cohort paper was checked, or a version was not recorded, the panel says so. Save a project to keep citations with the figure. See all references.
Some patients have no event during follow-up. n=100, e=30 means 30 events and 70 censored.
CPTAC has one outcome (OS). Adjusting one p-value leaves it the same.
Patients with missing expression, missing outcome or zero follow-up are left out. Custom percentile groups also leave out the middle.
Patients with equal values stay together.
No. Only RNA is included.
It may. Trying many cutoffs raises the chance of a false finding. Report what you tried.
The project, the JSON and the versions. Report the gene, cohort, outcome, grouping and excluded patients.