Identify Upregulated Genes From Ranks

Question

[Jessica]

Possible Duplicate:
Identify expressed genes from combined microarray data sets

Hi all,

I have a dataset consisted of 11 cell types with each cell type having >=5 replicates. For my biological question, I need to identify the unregulated (edit: should be up-regulated?) genes of each cell type and the genes do not have to be differentially expressed compared to all the others or cell type specific.

(1) I ranked all the genes within each array and would like to identify the genes that are consistently ranked in the top 50% of all the genes. What statistical methods that I shall use? How about estimating FDR? (2) To complement gene list from (1), I also need to find genes that are consistently ranked higher in the interested cell type than in all the other cell types, but these genes are not necessarily ranked within the top 50% of all the genes. What statistical methods that I shall use for this purpose?

Is there a BioConductor package to achieve these two goals.

Thank you very much. Any advice is much appreciated.

Wendy

[Michael]

I just saw that you have another open question about this. Please do not open duplicate questions.

[Sean Davis]

Hi, Wendy. This looks like another question you just asked?

Answer 1

[Michael]

Have a look at the RankProd package. It does meta-analysis based on ranks and is possibly closest to the approach you had in mind. Computing FDR is a good idea, it can be computed only after you have p-values though.