Skip to content

The 2D Plot and PCA

Analyze → 2D Plot opens a plot window for every selected data set. It is the main analysis window of CytoSpec: a scatter plot or histogram of any two channels - or of any two principal components - with gates.

The window has the chart and a docked control panel.

Choosing what to plot

The control panel has three lists:

List Purpose
Parameters The channels that take part in PCA. Select All / Unselect All tick or untick them
X What the horizontal axis shows: a channel, or a principal component once PCA has run
Y The same for the vertical axis

Tick one item in X only to get a histogram; one in each for a scatter plot.

Beside each axis list:

Control Effect
lin->log Show a linear channel on a log axis
log->lin Show a log-scaled channel on a linear axis

Redraw redraws the plot with the current choices.

Principal component analysis

PCA finds the directions in which the spectra vary most. Plotting the first two components often separates populations that no single pair of channels can.

  1. Tick the channels to use in the parameter list - at least two.
  2. Choose the matrix and options:
Control Meaning
Matrix type Correlation (every channel scaled to unit variance), Covariation (raw variances) or Cross-Products (not centred)
Normalize Make all vector lengths = 1 - divide every event's spectrum by its length, so only its shape counts
HyperSpherical Convert every spectrum to hyperspherical coordinates first - its length R and its angles PHI0, PHI1 ...
  1. Press PCA. The principal components are added to the X and Y lists, and the plot switches to the first two.

Show Tables opens the Eigenvalues, Eigenvectors and The Matrix tables; click a table to export it as CSV. Save PC and Load PC write and read the principal components, so the same projection can be applied to another data set.

The algorithm used to compute the eigenvectors is set in Properties.

Display

Control Effect
Dot Size Size of the points
Colour button (Paint Points) Colour a range of events
Density button Colour the points by local density
Show Labels Label the gates
Show Pointer Show a cross-hair pointer
Marker button (Show Markers) Show two vertical markers on a histogram
Preview button Push and Click any data point to preview - shows the spectrum of the event you click
Multi Show several data sets together

Point at a gate to see its name, its number of events and percentage, and the mean X and Y of its events.

Gates are described on the next page, Gates.