Skip to contents

Annotates cell populations found using CytomeTree.

Usage

Annotation(
  CytomeTreeObj,
  K2markers = NULL,
  K3markers = NULL,
  plot = TRUE,
  t = 0.2,
  remove_outliers_inplot = TRUE,
  center_fun = c("median", "mean")
)

Arguments

CytomeTreeObj

An object of class CytomeTree.

K2markers

A vector of class character where the names of the markers for which 2 levels of expression are sought can be specified. Default is NULL i.e. unsupervised.

K3markers

A vector of class character where the names of the markers for which 3 levels of expression are sought can be specified. Default is NULL i.e. unsupervised.

plot

A logical value indicating whether or not to plot the partitioning in 1, 2 or 3 groups for each marker. Default is TRUE.

t

A real positive-or-null number used for comparison with the normalized AIC computed to compare the fits of the marginal distributions obtained by one normal distribution and by a mixture of two or three normal. For markers used in the tree, the algorithm compares the fits obtained by a mixture of two and three normal distributions. Default value is .2. A higher value leads to a smaller number of expression levels per marker.

remove_outliers_inplot

a logical flag indicating whether the y-axis should be scaled by removing outliers or not. Default is TRUE.

center_fun

a character string either 'median' or 'mean' indicating based on which summary the populations should be ordered. Default is 'median', which is more robust to outliers and long tail distributions.

Value

A data.frame containing the annotation of each cell population.

Details

The algorithm is set to find the partitioning in 1, 2 or 3 groups of cell populations found using CytomeTree. In an unsupervised mode, it minimizes the within-leaves sum of squares of the observed values on each marker and computes the normalized AIC to compare the fits of the marginal distributions obtained by one normal distribution and by a mixture of two or three normal.For markers used in the tree, the algorithm compares the fits obtained by a mixture of two and three normal distributions.

Author

Chariff Alkhassim, Boris Hejblum