[BioC] linear model and continous variable

[James W. MacDonald]

Hi Dimitris,

On 6/30/2011 3:24 PM, Dimitris Kampas wrote:
> Hello everyone,
>
> I would like to perfom a linear model like the following on my data using
> limma but i am not familiar with that.
>
> Y~ a + b*TYPE + c*TIME + d * TYPE * TIME
> I have 2 types of mice, CSB and WT, and 3 time points.
> I would like to model the time as a continous covariate.
> Unfortunately, the limma examples i found contain only categrical
> covariates.
> Can anybody provide me any help? Or does anybody know where to find an
> example of limma model with
> continous covariate.

You don't say what kind of data you are analyzing, so I will keep this 
pretty general. What you need to understand is that the design matrix 
set up by R is dependent on the class of the input data. So if you pass 
numerical data to model.matrix(), then R will assume you want to model 
those data as a continuous variable. For example,

 > z <- 1:4

Fit z as a continuous variable:

 > model.matrix(~z)
   (Intercept) z
1           1 1
2           1 2
3           1 3
4           1 4
attr(,"assign")
[1] 0 1

Fit z as factor levels:

 > model.matrix(~factor(z))
   (Intercept) factor(z)2 factor(z)3 factor(z)4
1           1          0          0          0
2           1          1          0          0
3           1          0          1          0
4           1          0          0          1
attr(,"assign")
[1] 0 1 1 1
attr(,"contrasts")
attr(,"contrasts")$`factor(z)`
[1] "contr.treatment"


Does that help?

Best,

Jim


>
> Dimitris Kampas
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor

-- 
James W. MacDonald, M.S.
Biostatistician
Douglas Lab
University of Michigan
Department of Human Genetics
5912 Buhl
1241 E. Catherine St.
Ann Arbor MI 48109-5618
734-615-7826
**********************************************************
Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues