I have a microarray dataset compiled from several sources, so I am facing some challenges with identifying the expressed genes of each cell type. I am thinking to use the enriched gene sets of each cell type as the expressed genes of that cell type. However, the gene set enrichment analysis (http://www.broadinstitute.org/gsea/index.jsp) needs both control and sample data. I am wondering if there is gene set enrichment tool for the analysis of one cell type only.
I think refining the context of your problem may help you get specific answers. I am not clear what you mean by "microarray dataset" ? Do you have raw expression data or do you have information about differentially expressed genes ? Is your study designed to study differential expression of genes in multiple cell types ? Can you give some example of cell types that you are referring to ?
I think you already have the answer. You mention "I am thinking of using the enriched gene sets of each cell type...". OK. enriched compared to what? Probably some average gene expression you get from all your tissues, right? I suppose that average then would also be your control in the GSEA. Does that make sense?
Edit in reply to the comment and the other answer given
For any kind of enrichment analysis in sets of genes (for instance genes in specific pathways) you will calculate the number of genes in that pathway that are changed (or a quantitative measure for how much they are changed) relative to the whole pathway. In principle you would only need the list of changed genes for that. But if you really do that you do not take into account that you didn't actually measure all genes. Some genes were not present on your array for instance, or the array technology simply is not able to detect them. That is why you also want to know which genes are unchanged compared to a control sample and still measurable. You could of course assume that for tissue specific expression all genes were is your measured set, and assume that if you don't have an actual value they were not expressed. But it makes more sense to at least use the genes for which you don't have an expression signal as well.
In principle you could still do that for single tissues (so not use a kind of average for all tissues to calculate the enrichment), but then yo should not be surprised if some sets of genes (e.g. genes important for protein translation itself or say glycolysis) show up in all tissues.
Hi Chris, thank you for your suggestion. As my dataset is compiled from microarray data obtained using two platforms Illumina and Affymetrix). I am trying to avoid average the expression levels. I was wondering if there is gene set enrichment analysis for one cell type only without given a reference cell type.
Gene Set Enrichment Analysis works by ranking the "differentially" expressed genes between "two groups" and calculating a sum statistic eventually reporting a Enrichment score. As Chris pointed out if you ask a question like, "Is my set of genes are enriched in a particular cell type?", the implicit thing is you are comparing the gene set in your cell type of interest with "some" other.
As you said, if you want to find out whether your gene set is "enriched" (actually it would be misnomer to use for a single cell type), then you could just sort your list of genes based on the absolute expression values, then select a list of genes above a certain threshold (you could possibly calculate mean to guide you to select a threshold). Now you check how many genes in your list is present in the genes that are above threshold value.
lets say you have 10 genes in genelist, out of which 6 has been above your threshold value, then you could report 6/10 enriched in the cell type.
Two things you need to make sure,
(a) How do you combine the expression values of Affy and Illumina, wat strategy are you using to scale both the values in same range?
(b) What I suggested is just an idea, you have to extend that further
Current literature usually uses "Gene Set Enrichment analysis" when comparing two groups, so when we design a particular strategy specific for our analysis and trying to use the term "Enrichment", i thought that would cause confusion.
I think refining the context of your problem may help you get specific answers. I am not clear what you mean by "microarray dataset" ? Do you have raw expression data or do you have information about differentially expressed genes ? Is your study designed to study differential expression of genes in multiple cell types ? Can you give some example of cell types that you are referring to ?
Have a look here: http://www.bioconductor.org/packages/release/bioc/html/GSVA.html