Problem in POE transformation for dual channel microarray meta-analysis

Question

[greeshma_gt]

Hi,

I have been carrying out meta-analysis on few publicly available datasets for different platforms using MetaArray package in R. I have congregated all the raw files and normalized data from individual studies separately.

I tried to compute POE to obtain patterns of differential expression via poe.em() and got this error:

Error in while (err > threshold) { :
  missing value where TRUE/FALSE needed

There is bewilderment in calculating POE, could anyone help me to figure out on how to perform POE transformation for a normalized Agilent dual channel data and resolve the error.

I called the poe.em function with phenotype info as:

poe_dual_ch<-poe.em(na.omit(max_dual), dual_pheno$Cy5)

where max_dual is my expression data matrix and the pheno details from the file dual_pheno is below:

FileName    Cy3    Cy5
GSM_T2.txt    1    0
GSM_T3.txt    1    0
GSM_T4.txt    1    0
GSM_T5.txt    1    0

I also called poe.em with unknown phenotype option by sending a cl vector of zeroes:

poe_dual_ch<-poe.em(na.omit(max_dual[-1]), rep(0,4))

I've also tried poe.mcmc avoiding phenotype option resulting with the following error:

poe_dual_ch<-poe.mcmc(na.omit(max_dual[-1]))

Error in poe.mcmc(na.omit(max_dual[-1])) :
  NA/NaN/Inf in foreign function call (arg 4)

[David Westergaard]

How did you call the function? And did you read the help page (?poe.em)?

[greeshma_gt]

Yes, I've read the help page of poe.em. I've edited the question with function calls.