[BioC] expression set and paired designs

[David martin]

I have modified my pData.txt so that it looks like that:

Sample  Group celltype
sample1        moderate_disease celltype1
sample2        moderate_disease celltype2
sample3        severe_disease celltype1
sample4        severe_disease celltype2
sample5        normal	 celltype1
sample6        normal	 celltype2


following your suggestions:
My block factor is the Group created as follow:

#
#Creates design combinations
#Got it from the forum. Very helpfull
design.list <- function(phenotype,cond)
   {
n <- length(phenotype)
k = length(cond)
design <- matrix(1,1,(n))
for (i in 1:k)
   {
     #print(paste("Reading:",i))
     ind = (which(phenotype == cond[i]))
     design[,ind] = i

   }
design
   }


conditions=factor(c(design.list(phenotype$Group,groups)),labels=c("normal","moderate","severe"))
celltypes=factor(c(design.list(phenotype$Celltype,celltype)),labels=c("celltype1","celltype2"))


data<- exprs(mydata.eset)
fit<- manova(data ~ conditions * celltypes)
Error in model.frame.default(formula = data ~ conditions * celltype, 
drop.unused.levels = TRUE) :
   variable lengths differ (found for 'conditions')

Since i have different levels i can't compute the manova. Can you help 
me on that ?

The main question being, what is the effect of each celltype on disease 
and what is the effect of both celltypes on disease progression.

The first question can be resolved using classical anova but what about 
the seconf with two variables (celltype1 and celltype2)

thanks again for all the help i'm getting,




David martin wrote:
> Is there any simple example to start with ?
> 
> 
> Naomi Altman wrote:
>> What you have is a split plot design.
>>
>> The whole block factor is disease severity.  The blocks are patients.  
>> The subplot factor is cell type.
>>
>> Since there are only 2 cell types, you can readily determine the 
>> disease by cell type main effect, by analyzing the differences between 
>> the cell types for each
>> patient.  However, if the main effects are also of interest, you need 
>> to run the split plot design ANOVA for each gene.
>>
>> I am not sure whether you can do this in Limma, using patient as 
>> block.  If not, you should be able to do it in MAANOVA.
>>
>> --Naomi
>>
>> At 08:35 AM 12/7/2009, David martin wrote:
>>> Hi,
>>> Here is the experimental design (done by flow cytometry).
>>>
>>> Collect sample from a set of patients-> measure the expression for a 
>>> set of genes in different celltypes from the same sample.
>>>
>>> So the normalized data look like that:
>>>
>>>         celltype(1 or2) geneA   geneB   geneC
>>> patient1        1       40      20      40
>>> patient1        2       37      18      41
>>> patient2        1       40      19      38
>>> patient2        2       38      17      39
>>> patient3        1       10      19      38
>>> patient3        2       20      17      39
>>>
>>> ....(n)
>>>
>>>
>>> and then i have my pdata.txt.
>>>
>>> Sample  Disease_stage
>>> patient1        moderate_disease
>>> patient2        severe_disease
>>> patient3        normal
>>>
>>>
>>> What i want to do is to compare the different groups and identify the 
>>> genes that differentially expressed between the three groups.
>>> That i guess would be fine to do by bulding a proper design and 
>>> runing a paired t.test.
>>>
>>>
>>> But before that I can't construct an eset object as sample names are 
>>> duplicates. I was wondering if i need to construct two eset objects 
>>> (one for celltype1 and one for celltype2) ???
>>>
>>> Any help would be appreciated.
>>>
>>> thanks
>>>
>>> _______________________________________________
>>> Bioconductor mailing list
>>> Bioconductor at stat.math.ethz.ch
>>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>>> Search the archives: 
>>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>
>> Naomi S. Altman                                814-865-3791 (voice)
>> Associate Professor
>> Dept. of Statistics                              814-863-7114 (fax)
>> Penn State University                         814-865-1348 (Statistics)
>> University Park, PA 16802-2111
>>
>> _______________________________________________
>> Bioconductor mailing list
>> Bioconductor at stat.math.ethz.ch
>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>> Search the archives: 
>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: 
> http://news.gmane.org/gmane.science.biology.informatics.conductor
>