Chart components

Chart nodes have purple title bars and live under + Visual. They all take a dataset, most take a filter and a marking, and they all report a selection.

Common ground

Fields, on the node

A ScatterPlot node with dropdowns for X axis, Y axis, Colour, Size and Shape, input ports on the left and a selection output on the right.

Each chart’s fields are chosen on its node in the Design tab. Only fields of a suitable type are offered, and a field marked * is required.

Selecting

Drag a box over a chart to select what is inside it. The selection is drawn distinctly — rings around selected points in a scatter plot, a solid bar against a pale one in a histogram — and the Selected count in the status bar says how many rows it holds. Clicking empty space clears it.

Two charts select whole bins rather than individual rows, because that is what they draw: the Histogram outlines a whole bar holding a selected row, and the HexbinPlot outlines a whole hexagon. For a finer grain, use smaller bins.

Settings

The Appearance tab of a scatter plot’s settings: base colour, colour scale, then a linear/log toggle and an ascending/descending toggle for the colour, then point size.

What applies to every component’s ⚙ — tabs, auto (…) placeholders, the Tooltip tab, and the fact that settings belong to the view — is under Settings. What the charts share on top of that is below.

Base colour is what the chart draws in when no field is mapped to colour. Until you set one, the chart follows the application’s default colour; the ↺ beside it puts the chart back onto that default. Colour scale is the ramp used when a field is mapped to colour; Default (…) follows the application’s default colour scale, and names the one it currently is.

Wherever a numeric field is mapped to something visual — a colour, a size, an axis — two toggles follow it:

Linear / Log — the scale. Log is for values spanning orders of magnitude, such as an IC50. It cannot show zero or negative values, and rows carrying them are counted as out of range.
Asc / Desc — which end of the field counts as the top of the scale.

Aggregation appears on the three charts where several rows share one mark — the histogram, the integer scatter plot and parallel coordinates. It decides how their colour values are combined: Mean, or Geometric mean for a quantity that is naturally multiplicative.

Large datasets

A chart that draws a mark per row — the ScatterPlot, FacetScatterPlot, MultiFieldScatterPlot and ParallelCoords — draws at most Max points of them (10,000 unless you change it; the box shows auto (10000)). Past that it draws an evenly spread sample, the same rows every time, and says so in the bottom left of the chart: Showing a sample of 10,000 of 1,000,000 points. Rows you have marked are always drawn.

The sample only limits what is drawn. A selection still covers every row in the region you select, not just the ones on screen, and axes still span the whole dataset. Raise Max points, or narrow the rows with a filter, to draw more. The Histogram, HexbinPlot, IntegerScatterPlot and BinnedBoxPlot summarise all rows into bins and have no such limit.

ScatterPlot

X against Y, one point per row, with optional colour, size and shape fields. The workhorse.

A scatter plot of lipophilicity against polar surface area, coloured by heavy-atom count, with a box selection ringed in magenta and a colour legend beneath.

Fields: X axis*, Y axis*, Colour, Size, Shape.

Axes | Appearance | Tooltip

The Axes tab: linear/log and ascending/descending toggles for each axis, and minimum and maximum boxes showing the automatic values.

Tab	Settings
Axes	X and Y: scale type, direction, minimum, maximum
Appearance	Base colour, colour scale, colour scale type and direction, point size — plus size scale type and direction when a size field is mapped — and Max points
Tooltip	The tooltip expression

FacetScatterPlot

The same plot split into a grid of small multiples by two further fields — one choosing the row, one the column. Use it when a relationship might differ between groups: the grid shows all the groups at once, on shared axes, so the comparison is direct.

A facet scatter plot: a grid of small scatter plots of lipophilicity against polar surface area, split by ring count down the side and aromatic ring count across the top.

Fields: X axis*, Y axis*, Row facet*, Col facet*, Colour by.

Selection works per facet — drag a box inside one cell. It shares the ScatterPlot settings panel, minus the point-size and size-field rows.

MultiFieldScatterPlot

A grid of scatter plots, one for every combination of the X fields and the Y fields you choose. Dragging a box selects across all of them at once.

A multi-field scatter plot: a two-by-two grid of scatter plots, each pairing one of two X fields with one of two Y fields.

Fields: X fields* (two or more), Y fields* (two or more), Colour by.

Appearance | Tooltip

The Appearance tab: multi-select logic, base colour, colour scale, point size and shared axis limits.

Multi-select decides how selections from several subplots combine: OR — union takes anything matched anywhere, AND — intersection only what every subplot agrees on. The rest of the tab is base colour, colour scale, point size and a shared minimum and maximum for X and Y.

With many independent axes on screen, this chart has no per-axis scale or direction controls.

IntegerScatterPlot

For fields with a handful of distinct whole-number values — ring count against rotatable bonds, say. Rows landing on the same pair are drawn as one circle, sized by how many, so an ordinary scatter plot’s overplotting problem disappears.

An integer scatter plot of ring count against rotatable bonds, each position a circle sized by how many compounds share it and coloured by mean lipophilicity.

Fields: X axis*, Y axis*, Colour.

A flat panel: base colour, colour scale, aggregation, colour scale type and direction, then a minimum, maximum and direction for each axis.

The integer scatter plot’s settings: base colour, colour scale, aggregation, colour scale type and direction, and axis limits and directions.

The axes deliberately have no Linear/Log toggle: counts routinely include zero, where a log scale has nothing to show.

It makes one circle per distinct pair, so it is for discrete fields. A field with more than 100,000 distinct positions (a continuous measurement, say) is refused with a message saying so: use a ScatterPlot or HexbinPlot for that. The tooltip gives the count at a position, not the individual rows.

Histogram

The distribution of one numeric field.

A histogram of polar surface area, with the bars belonging to an upstream selection drawn solid against the pale full distribution.

Fields: Field*, Colour.

A flat panel: base colour, colour scale and aggregation, then Bin start, Bin step and Bin count. Leave them blank and the bins are worked out from the data; the placeholder tells you what that came to. A histogram has at most 1,000 bars, so a larger Bin count is treated as 1,000.

The histogram’s settings: base colour, colour scale, aggregation, and bin start, step and count.

A histogram is the natural target for a marking connection: the whole distribution stays on screen and your selection is drawn against it, which is exactly the comparison you want.

ParallelCoords

One vertical axis per field, one line per row crossing them all — the way to see how several properties move together across a series, and where the trade-offs are.

A parallel coordinates plot of five properties, a thousand lines coloured by lipophilicity.

Fields: Fields* (two or more), Colour.

A flat panel: base colour, colour scale, aggregation, the orientation of the axes, and Max points (shown as lines in the caption).

The parallel coordinates settings: base colour, colour scale, aggregation and orientation.

BinnedBoxPlot

The distribution of one field within equal-width bins of another — median, quartiles, whiskers and outliers per bin. A scatter plot shows you every point; this shows you the shape of the trend.

A binned box plot of polar surface area against bins of lipophilicity, each bin drawn as a box with whiskers and outliers.

Fields: X axis* (binned), Y axis* (distribution), Colour.

Binning | Axes | Tooltip

The Binning tab: orientation, bin start, bin width and bin count.

Tab	Settings
Binning	Orientation, bin start, bin width, bin count
Axes	Distribution axis scale type and direction, minimum and maximum; base colour and colour scale, and the colour scale type and direction
Tooltip	The tooltip expression

The chart draws at most 20,000 outlier points; if there are more, a caption says how many there were in all.

HexbinPlot

A density map. Points are gathered into hexagons and each hexagon is coloured by how many rows fell into it — or by an aggregate of a field across them.

A hexbin plot of lipophilicity against polar surface area, the densest hexagons brightest, with a count legend.

Fields: X field*, Y field*.

Axes | Appearance | Tooltip

The hexbin Appearance tab: colour scale, what colour encodes, and bin start, size and count.

Colour by decides what a hexagon’s colour means: count, or the mean, sum or median of a field you then choose. Bin size (px) sets how coarse the grid is.

Use it when a scatter plot has too many points to read — tens of thousands of rows overplot into a solid blob, and a hexbin says where the mass actually is. It takes a filter but no marking: it draws bins, not rows. A hexagon with no value for the field you colour by is left uncoloured.