Tweet
Having spent a fair amount of time on learning the basics of RNA-seq and its data analyses, I'm now trying to learn about GO enrichment of my results. I have two cases for which I'd like to do some kind of GO enrichment analysis in human:
1) Simple gene lists containing only IDs (ENSGID in my case) of genes that are considered to be expressed (FPKM > 1) in various cell types (cancer/normal/cell lines) and the uniquely expressed genes between the types. For example, 100 genes only expressed in the cancer cell data: what GO-terms are enriched, if any?
2) Lists of DESeq results, i.e. differentially/non-differentially expressed genes with p-values, FDR, fold change, etc. between the same groups above. Among the DE-genes, what GO-terms are expressed?
I've been trying to read up on the honestly humongous amount of various softwares (web-based and not) that seem to be out there and what type of analyses people tend to do. Case (1) seems to be the simplest, and case (2) can of course be converted to (1) by just taking the DE-genes' IDs into a separate list. Is case (1) a common to thing do? It seems to be (a beginner) that that sort of thing would be very useful to do on a wide range of different experimental data, but does it yield useful/trustworthy results?
For (2), is this something people do, i.e. having some p-values and/or FDR-values along with the IDs? What does this say that is different from (1) - does it give more robust statistical results, perhaps?
Now, I've read that a lot of people use DAVID, which I tried more or less successfully for my data sets. The "Functional annotation clustering" seems to be the kind of result that I'm after, but what do people usually look at? I also tried WebGestalt, which seem to give a more visual result. I could not get GOstat to work at all (outputs 0 enriched genes), for some reason. These are all web-based tools; what are some programmatical tools/modules, preferably for Python (or R) that I could try once I'm feeling more comfortable with the analysis in general?
I'm basically in the beginning and I'm wondering what tool(s) I should delve deeper into based on what kind of results I'm looking for, and would appreciate tips. Also, I'd be good to know what type of GO-studies people generally do with the type of data I have (i.e. various expression data from RNA-seq).
Thanks in advance!
1) Simple gene lists containing only IDs (ENSGID in my case) of genes that are considered to be expressed (FPKM > 1) in various cell types (cancer/normal/cell lines) and the uniquely expressed genes between the types. For example, 100 genes only expressed in the cancer cell data: what GO-terms are enriched, if any?
2) Lists of DESeq results, i.e. differentially/non-differentially expressed genes with p-values, FDR, fold change, etc. between the same groups above. Among the DE-genes, what GO-terms are expressed?
I've been trying to read up on the honestly humongous amount of various softwares (web-based and not) that seem to be out there and what type of analyses people tend to do. Case (1) seems to be the simplest, and case (2) can of course be converted to (1) by just taking the DE-genes' IDs into a separate list. Is case (1) a common to thing do? It seems to be (a beginner) that that sort of thing would be very useful to do on a wide range of different experimental data, but does it yield useful/trustworthy results?
For (2), is this something people do, i.e. having some p-values and/or FDR-values along with the IDs? What does this say that is different from (1) - does it give more robust statistical results, perhaps?
Now, I've read that a lot of people use DAVID, which I tried more or less successfully for my data sets. The "Functional annotation clustering" seems to be the kind of result that I'm after, but what do people usually look at? I also tried WebGestalt, which seem to give a more visual result. I could not get GOstat to work at all (outputs 0 enriched genes), for some reason. These are all web-based tools; what are some programmatical tools/modules, preferably for Python (or R) that I could try once I'm feeling more comfortable with the analysis in general?
I'm basically in the beginning and I'm wondering what tool(s) I should delve deeper into based on what kind of results I'm looking for, and would appreciate tips. Also, I'd be good to know what type of GO-studies people generally do with the type of data I have (i.e. various expression data from RNA-seq).
Thanks in advance!