Skip to content

Compensation

Compensation corrects fluorescence spillover with a square spillover matrix: one row and one column per fluorescence detector, each off-diagonal entry saying how much of the row's dye lands in the column's detector.

PlatAnJr's compensation editor is the Parameters tab in the bottom-left info pages. It is always open - there is no separate dialog to launch.

Compensation or unmixing?

Compensation inverts a square detector-by-dye matrix and suits conventional filter-based cytometers. For a spectral instrument with more detectors than dyes, use Spectral unmixing instead: it solves an over-determined system and gives you a choice of solvers and regularisation.

The editor

Area What it is
Parameter list (left) Every parameter, each with a checkbox marking it as a log parameter
Scale type and Decades / negative decades The default display scale applied to the checked parameters
Crop Outliers Clamps extreme values instead of letting them stretch the scales
Spillover grid (right) The matrix itself - editable cell by cell, with parameter names down the side and across the top
Use Spillover matrix Master switch: apply compensation or not
Normalize See below
Apply Commit everything on the page and recompute
Dismiss Leave the page without committing
Import / Export Read a matrix in from a CSV file, or write the grid out as one - see Loading a matrix

Loading a matrix

Most instruments write an acquisition spillover matrix into the FCS keywords, and PlatAnJr reads it when the plate is opened - $SPILLOVER, BD's SPILL or Partec's SPILOVER, whether it lists the parameters by name or, as older CyAn and IntelliCyt files do, by number. The grid is filled in for you, and Use Spillover matrix turns it on and off.

To bring in a matrix from elsewhere, click Import and pick a CSV file. Its rows and columns are matched to the loaded parameters by name - the parameter name ($PnN), the name the grid shows, or the stain ($PnS) - whatever their order, so a matrix exported with every channel can be imported onto a plate opened with only some of them. It reads PlatAnJr's own exports, R's write.csv, and spreadsheets saved with ; and decimal commas. The imported matrix replaces the grid; click Apply to use it, also on data that came with no matrix of its own.

Export writes the current grid out as CSV, under the parameter names and with . as the decimal separator - useful for sharing a matrix or for editing it in a spreadsheet. It is greyed out while the grid is empty.

The grid is directly editable: click a cell, type a new coefficient, and click Apply.

Opening only some parameters

When the open dialog's parameter picker leaves out some parameters, the matrix keeps the ones that are loaded. Compensating a channel needs the readings of every detector whose dye spills into it, though: if a channel left out - or one the matrix names that is not in the file - spills into the loaded ones, PlatAnJr names it in red under the grid, and that spillover stays in the loaded channels. Load those channels too to compensate fully. Leaving out a channel whose dye spills into none of the loaded ones loses nothing.

The Normalize checkbox

The diagonal of a spillover matrix should be 1 - a dye contributes 100% of itself to its own detector. Matrices from other software are not always written that way.

  • Normalize ticked - the matrix is used exactly as it appears in the grid.
  • Normalize cleared - each row is divided by its own diagonal element before use, so the diagonal reads 1 whatever the absolute numbers in the file were.

If an imported matrix gives obviously wrong results and its diagonal is not 1, clear Normalize and apply again.

Applying

Click Apply. PlatAnJr recomputes the compensated values and tells the rest of the application that the data changed; every chart node, gate and statistic updates.

Gates are defined in scaled display space, so a change in compensation moves the data under your gates rather than moving the gates. Expect to revisit gate positions after changing a matrix, and re-check the populations at the top of your gating tree first.

Detector linearisation (Straiten)

The Straiten button applies a one-off linearity correction to analog and photon-counting channels, using the sensor calibration data on Info ▸ Calibration.

  1. Click Straiten.
  2. Check the parameters to be corrected.
  3. Click OK.

Channels whose names end in -AA (analog) and -AP (photon count) are mapped through their calibration curves onto a common linear scale.

This can only be done once per loaded dataset - a second attempt reports "Straightened Already". If you need to redo it, re-open the plate.

Which values are compensated

Everything downstream works on compensated values:

  • charts and gates,
  • statistics,
  • derived parameters, which are computed from the compensated data,
  • the heat map.

Turning Use Spillover matrix off does not lose the matrix; it just stops applying it, so you can toggle between raw and compensated views to judge whether the matrix is doing what you expect.