Plots (Bokeh)
Plots (Bokeh) is the second plotter in Mantis. The chart is built on the backend with Bokeh and embedded in the panel, which makes it slower to update than Plot Controls (Simple) but gives you plot types the simple plotter does not have: box plots, violin plots, and a correlation heatmap across every numeric column at once.
Unlike the simple plotter, this panel keeps several plots at a time as tabs, and it is self-contained: controls on the left, canvas on the right.
Where to find it
View → Quantitative → Plots (Bokeh). You can also right-click the map with two or more points selected and pick Visualize Selection under Visualize.
What it plots
The bag picker, Select Bags/Selections, takes any number of bags. Leave it empty and it reads All Points in the space instead. It preselects the most recently created bag when the panel opens.
The feature list is every metadata column with numeric values across the chosen points, plus Embedding X and Embedding Y for the projection coordinates.
Controls
The form is numbered in the panel:
- Choose Analysis Type. Analyze Selection looks inside one set of points; Compare Selections puts several bags side by side. The choice decides which plot types are available.
- Plot Type.
- Under Analyze Selection: Histogram, Scatter Plot, Correlation Heatmap.
- Under Compare Selections: Box Plot, Violin Plot.
- Select Data. The bag picker described above.
- Select Features. Feature (X-Axis), plus Feature (Y-Axis) for scatter plots. The Correlation Heatmap skips this step: it uses every numeric feature. Under each dropdown a Field stats card reports n, missing, min, mean, max, and standard deviation for the fields you picked, including the color and size fields.
- Plot Configuration (Optional). Title, X-Axis Label, Y-Axis Label, and Log Scale (X-Axis); scatter plots add Log Scale (Y-Axis). Leave the title blank and one is generated, for example
Histogram of length.
Scatter plots add two collapsible sections:
- Color Configuration: Fixed with a color picker, By Feature with a numeric column, or By Selection to color each bag differently.
- Size Configuration: Fixed with a marker size, or By Feature with a numeric column.
Create Plot renders the chart. Compress outliers, beside the button, clamps the axis range instead of letting extreme values flatten everything else. The button stays disabled until the panel has data and a feature: Compare Selections needs at least one bag checked, and every type except the heatmap needs an X feature.
Managing plots
The + button in the panel header adds a new empty plot; Rename Plot names the current one. Each plot gets a tab above the canvas with a close button, and Active Plot in the header switches between them. Configuring and pressing Create Plot replaces the contents of the active plot rather than adding a new one.
Selection is wired back to the map
Histogram bars and scatter points carry their point ids. Selecting them in the Bokeh chart makes those points the current Mantis selection, so the Inspector and the other panels follow. A single selection carries at most 1,000 points; anything beyond that is dropped.
How to use it
- Open the panel and pick an analysis type.
- Choose bags. Leave the picker empty to plot the whole space.
- Pick the plot type and features, then press Create Plot.
- Select in the chart to push those points back into the Mantis selection.
- Press + to start a second plot and compare configurations across tabs.
Limits
- Nothing renders until you press Create Plot. This panel does not follow the selection live the way the simple plotter does.
- Box and violin plots require Compare Selections and at least one checked bag; with no bag checked there is nothing to compare.
- The Correlation Heatmap is computed in the browser over the numeric features of the chosen points, and needs at least two of them.
- Plots are not saved with the space. Closing the panel loses them.
- Backend errors are shown in the canvas verbatim, prefixed “Backend Error”.