Hi everyone,

I have two geneset (microarrays) from the same type of tumor. The platform is slightly different, I both normalized with z-score and I did the correlation matrix (ex MYC with all the other genes present in each dataset separetly) I would like to correlate 1 gene with all the others in the two dataset at the same time. To make an example:

geneset1 <- data.frame(genename = seq(1,10,1), replicate(10, rnorm(10,1,2)))
geneset2 <- data.frame(genename = seq(1,12,1), replicate(5, rnorm(12,1,2)))

After normalizing them, I did a correlation gene list in both dataset with cor function R. for example:

corr_matrix <- data.frame(correlation = sapply(1:nrow(geneset1), 
                                           function(x) cor(t(geneset1[1,]), t(geneset1[x,]), 
                                                           method = "pearson", 
                                                           use = 'pairwise.complete.obs') ),
                      pValue = sapply(1:nrow(geneset1), 
                                      function(x) cor.test(t(geneset1[1,]), t(geneset1[x,]), 
                                                           method = "pearson",  use = 'pairwise.complete.obs')$p.value)

but now I want to join the datasets and to have a unique correlation gene list. what do you suggest to obtain a common list of correlation among one gene and the others ?

• To do two separate lists and to do a weigth correlation for each gene? ex: (CorGeneset1 * Nsample1 + CorGeneset2* Nsample2) / (Nsample1 + Nsample2)

• or you suggest to join the dataset, to do after a normalization of the joined dataset, and then to do another correlation matrix?

• moreover, one dataset has some probes for genes not present in the other. What do you suggest to do?

Thank you everyone for your help