Skip to contents

Colour-coded scatter plots of e.g. LLR(PO/U) against LLR(FS/U), for various relationship combinations.

Usage

PlotPairLL(
  PairLL,
  combo = list(c("FS", "PO"), c("HS", "FS"), c("GP", "HS"), c("FA", "HS")),
  nrows = NULL,
  ncols = NULL,
  bgcol = TRUE,
  Tassign = 0.5,
  Tfilter = -2
)

Arguments

PairLL

dataframe, output from CalcPairLL.

combo

list with length-2 character vectors, specifying which likelihoods to plot against each other. Choose from 'PO', 'FS', 'HS', 'GP', 'FA', and 'HA'. The first one gets plotted on the x-axis, the second on the y-axis. Subsequent figures will be drawn row-wise.

nrows

number of rows in the figure layout. If NULL, set to ceiling(length(combo)/ncols).

ncols

number of columns in the figure layout. If both nrows and ncols are NULL, ncols is set to ceiling(sqrt(length(combo))), and nrows will be equal to ncols or one less.

bgcol

logical, colour the upper and lower triangle background of each figure to match the specified relationship combo.

Tassign

assignment threshold, shown as grey square in bottom-left corner and a band along the diagonal.

Tfilter

filter threshold, shown as dark grey square in bottom-left.

Details

The colour of each point is determined by columns focal (outer circle) and TopRel (inner filling) of PairLL.

Impossible relationships (LL > 0 in PairLL) are shown as -Inf on the axes, if any are present.

See also

Examples

Pairs <- data.frame(ID1 = "a01005",
                    ID2 = c("a00013", "a00008", "a00011", "b00001",
                            "b01006", "b01007", "b01013", "b01014"),
                    focal = rep(c("PO", "HS"), each=4))
PLL <- CalcPairLL(Pairs, GenoM=SimGeno_example, Plot=FALSE)
#>  Not conditioning on any pedigree
PlotPairLL(PLL,
           combo = list(c("FS", "PO"), c("HS", "FS"), c("GP", "HS"),
                        c("FA", "HS"), c("HA", "FA"), c("FA", "GP")),
           nrows = 3)