sumstats {made4} | R Documentation |
Given a data.frame
or
matrix
containing xy coordinates, it returns the slope and distance from
origin of each coordinate.
sumstats(array, xax = 1, yax = 2)
array |
A data.frame or
matrix containing xy coordinates, normally a $co, $li
from dudi such as PCA or COA, or $ls from
bga |
xax |
Numeric, an integer indicating the column of the x axis coordinates. Default xax=1 |
yax |
Numeric, an integer indicating the column of the x axis coordinates. Default xax=2 |
In PCA or COA, the variables (upregulated genes) that are most associated with a case (microarray sample), are those that are projected in the same direction from the origin.
Variables or cases that have a greater contribution to the variance in the data are projected further from the origin in PCA. Equally variables and cases with the strong association have a high chi-square value, and are projected with greater distance from the origin in COA, See a description from Culhane et al., 2002 for more details.
Although the projection of co-ordinates are best visualised on an xy plot, sumstats
returns the slope and distance from origin of each x,y coordinate in a matrix.
A matrix (ncol=3) containing
slope angle (in degrees) distance from origin
Aedin Culhane
data(khan) if (require(ade4, quiet = TRUE)) {khan.bga<-bga(khan$train[1:1000,], khan$train.classes)} s.var(khan.bga$bet$ls, col=as.numeric(khan$train.classes), clabel=0.8) st.out<-sumstats(khan.bga$bet$ls) # Get stats on classes EWS and BL EWS<-khan$train.classes==levels(khan$train.classes)[1] st.out[EWS,] BL<-khan$train.classes==levels(khan$train.classes)[2] st.out[BL,] # Add dashed line to plot to highlight min and max slopes of class BL slope.BL.min<-min(st.out[BL,1]) slope.BL.max<-max(st.out[BL,1]) abline(c(0,slope.BL.min), col="red", lty=5) abline(c(0,slope.BL.max), col="red", lty=5)