clin2mim {SAGx} | R Documentation |
Given a clinical variable, it produces a script file for WinMIM by calculating means and covariances and for the N most highly correlated probes (in absolute value). Here N is an input parameter, but a recommended value 10. WinMIM can find a relevant graphical model for the dependencies between the probes and the clinical variable.
clin2mim(variable="FEV1.ACTUAL",data=dbs,clindat=clinical,probes=probes,N=10,out="mimscr.txt")
variable |
Clinical variable to be examined |
data |
The input data set, with subject id in first column. |
clindat |
The input clinical data, with subject id in first column |
probes |
The name of the probes in the order of {it data} |
N |
|
out |
The MIM script file |
The correlation matrix
David Edwards' program WinMIM can be found on
StatLib (http://lib.stat.cmu.edu/graphmod/). In MIM issue input mimscript.txt
and the calculations to find a model will start. When finished go to the Graphics menu and
click on Independence Graph
. The resulting graph can be exported both to WMF and LaTeX.
Per Broberg
Edwards, David (1995) Introduction to Graphical Modelling. Springer-Verlag
Lautitzen, Steffen (1996) Graphical Models. Oxford University Press
Whittaker, Joe (1990) Graphical Models in Multivariate Analysis. Wiley