Parameters and scales¶
Every parameter in a dataset carries a scale: the transform that maps raw data values to the positions you see on an axis. Scales are a property of the parameter, not of a particular plot, so setting one changes every chart that plots that parameter.
The eight scale types¶
| Name | What it does | Adjustable parameter |
|---|---|---|
| Linear | No transform | - |
| Log | Base-10 logarithm, with a floor a fixed number of decades below 1 | Negative decades |
| Bi-o-Log | Logarithm of the value plus one, so zero and small values stay on scale | Negative decades |
| LinLog | Linear through a band around zero, logarithmic outside it | Linear width |
| ExpLog | Exponential-to-logarithmic blend | Linear width |
| Log^2 | Squared-logarithm scale | Linear width |
| Hyperlog | The hyperlog display scale | Linear width |
| BiExp | The biexponential (logicle) display scale | Linear width, in decades |
Linear is the only type with no adjustable parameter. All the others have exactly one knob, and that knob is what the scale bar on a chart drives.
Which one to use¶
- Linear for scatter (FSC/SSC) and for anything already on a linear detector scale.
- Log for classic log-amplified data with no meaningful negatives.
- BiExp (logicle) is the usual choice for compensated or unmixed fluorescence: it reads as log10 in the top decades, stays finite and symmetric through zero, and does not pile compensated negatives onto the axis. Its Linear width is measured in decades of the display scale.
- Hyperlog, LinLog and ExpLog solve the same problem with different curves; pick whichever gives your populations the shape you expect.
The scale bar¶
When a chart plots a parameter whose scale has an adjustable parameter, the chart grows a track bar along the corresponding axis - along the bottom for X, down the left for Y. Its tooltip reads out the axis, the scale type, the parameter name and the current value, for example:
X BiExp - Linear width: 0.65 dec
Drag the thumb and the picture stretches live. The value is committed on release, not while you drag: only then is the new scale applied to the parameter and broadcast to the other nodes, which redraw with it.
A chart too small to show a bar does not get one. The bar disappears if you switch the axis to a scale type with no adjustable parameter.
Changing the scale type¶
Double-click any chart to open the chart editor. It has an X Scale and a Y Scale combo listing the eight types. Choosing one rebuilds that axis's scale for the new type over the same data range, re-hangs the axis converters, refreshes the track bar and tells everything else that the scale changed.
The editor also carries:
| Control | Effect |
|---|---|
| Autoscale | Fit both axes to the data now |
| X SuperMax / Y SuperMax | Autoscale only ever widens the axis to cover new data, never shrinks it. Untick and re-tick to start over |
| Density Plot Style | Colour map or contour |
| Dot Size | Marker size on scatter plots |
| Show Grid, Show Legend, Background Color | Chart appearance |
| Copy To Clipboard, Export as CSV | Take the picture or the numbers out |
SuperMax is the setting to reach for when a plot flickers between scales as gates change: it pins the widest window the data has ever needed.
Setting a scale for the whole project¶
There is no central scale editor: a scale is set wherever the parameter is plotted, and the change propagates. Adjust the width bar or the scale type on any chart node showing that parameter, and every other node showing it redraws with the new scale.
The practical routine is therefore:
- Put the parameter on a chart node and get the population looking right there.
- Everything else that plots it follows.
- Save the project - the scales go with it.
The Parameter Scales page¶
Options ▸ Parameter Scales shows a table of every parameter:
| Column | Meaning |
|---|---|
| Parameters from File | The name as written in the FCS file |
| Parameters from Database | The resolved name, from the parameter database, if there is one |
| Minimum / Maximum | The data range PlatAnJr derived the scale over |
Click a row and the two charts beside the table draw that parameter's transform and its inverse, over its own data range. This is the place to look when an axis behaves strangely - a scale whose transform is not monotonic, or is flat over most of the range, will be visibly wrong here.
Parameter info¶
Info ▸ Parameter Info lists the per-parameter metadata PlatAnJr derived - a row for every parameter of the dataset, those added since it was opened included - with Export Table To CSV. Info ▸ FCS Info shows the raw FCS keywords of the selected file, which is the authority when a name, range or amplification setting looks wrong.
Picking several parameters¶
Where PlatAnJr asks for a set of parameters - Select Pars on a node, the markers of a phenotyping node, and elsewhere - it opens a picker with two lists, Dismissed and Selected:
| Control | Effect |
|---|---|
| >>>>>>> / <<<<<<< | Move the highlighted entries across - every entry when none is highlighted |
| <<Scatter | Take scatter (FSC, SSC) and Time off the selected list, however the instrument spells them |
| Ok / Cancel | Accept the selection, or leave it as it was |
Scales and projects¶
The set of scales a project uses is saved with the project, keyed by parameter name. When you open a project:
- parameters whose scale is unchanged are left alone;
- parameters whose scale differs are re-transformed.
That is why reopening a project is fast even on a large plate. It also means a project
carries its display conventions with it: send someone the .ppj and they see your axes,
not theirs.