[BioC] technical + biological replicate design for multiple categories

Naomi Altman naomi at stat.psu.edu
Thu Jul 4 03:45:06 CEST 2013



The blocking variable can have text.  Your treatment variable is now 
correct and as far as I can see, your analysis should now be fine.

--Naomi

At 05:34 PM 7/3/2013, you wrote:
>On Jul 3, 2013, at 1:32 PM, Naomi Altman wrote:
>
> > You need to have a variable "biorep" with biorep number e.g. 
> c(1,1,2,2,3,3,...6,6)
> >
> > This is a blocking variable.
>
>So I have to use numbers for the blocking variable? I can't use 
>text?  Otherwise, it's the same as the variable which I called "block."
>
> > The treatments are just Cyt1, Cyt2 and Mock.
>
>My design matrix should have just those three columns?  Then, my 
>variable "treatment" should actually be the following:
>
>treatment <- c("Mock", "Mock", "Cyt1", "Cyt1", "Cyt2", "Cyt2", 
>"Mock", "Mock", "Cyt1", "Cyt1", "Cyt2", "Cyt2")
>
>
> > You will need to follow the instructions for blocking variables.
>
>If I understand the user guide correctly then it's the same as what 
>I did before except for having the corrected "biorep" and 
>"treatment" variables:
>
>design = model.matrix(~ 0 + treatment)
>dupcor = duplicateCorrelation(expr, design, block=biorep)
>fit.raw = lmFit(expr, design, block=biorep, correlation=dupcor$consensus)
>fit = eBayes(fit.raw)
>
>This runs without error but I just want to make sure that it's 
>correct.  The results do make sense.  In other words, the top ranked 
>genes include many that were expected.
>
>Thanks,
>James
>_______________________________________________
>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



More information about the Bioconductor mailing list