[BioC] targets frame

FRANKLIN JOHNSON [guest] guest at bioconductor.org
Sun Jun 30 04:46:05 CEST 2013


Dear Maintainer,

I'm analyzing an expression matrix using a single factor with only two levels, 0 and 1.
I would like to add in a more than two factor variables from the experiment which measured percent disease progression, etc.
Injury development occurs 0, 1, 2, 3, 4 over time in the control and 0, 1, 2, 1, 1, (i.e. recovery) for the mutant over time of 0, 2, 4, 8, 12 week sampling points in 2011, and 0, 24hr, 72hr, 1, 2, 4 wks in 2012(or, second round of study).

My current setup is control.timepoint and mutant.timepoint for each TARGET in the targets.frame for comparison using the contrast.matrix. I've made a model using this setup as indicated in the limma.man.    

I hope to now add the development factor, etc into the model using limma through the design matrix. Currently, 
f=factor(targets$TARGET, lvls=TARGETnames) 
design = model.matrix(~0+f) 

To add other factors, it seems desirable to break up the current X.timepoint for each TARGET into other variables to help to contrast 0 and 1. For example, TARGET=c(rep(X1, X2), each =3)), then have TIMEPOINT/ %DISEASEPROGRESSIONRATE/OBSERVATIONPOINT/...

However, it's just as simple to name each array in TARGETS as "X.timepoint.%DPR.OP", keeping the matrix to a single factor setup with an 'intuitive name'?; and not creating variants of a 'full'design=model.matrix(~0+tp*%DPR*OP), to mine the targets using limma as well as with msm R package located here:
http://rss.acs.unt.edu/Rdoc/library/msm/doc/msm-manual.pdf  
 
What does limma.man suggest for the contents of this email?
 
Regards,
Franklin



 -- output of sessionInfo(): 

This is not a limma package, source code technical issue.
This is a limma application issue. 
#Thank you.  

--
Sent via the guest posting facility at bioconductor.org.



More information about the Bioconductor mailing list