Hi all,
I have several RNA-seq data(all of them have replication),after RNA-seq analysis,I want to perform met-analysis (for given a set of genes that are up-regulated or down regulate under drug conditions), I am wondering the best way to normalize the data - just calculate RPKM values or I should perform some sort of upper normalization? If so, what is the best way to do this?
1 answer
Use the R packages DESeq2 or EdgeR and let the packages do the normalization for you; it is the easiest approach and they are based on tested methods. RPKM is generally not an acceptable measure to use in current methodology; the statistics have come a long way since that calculation was developed. Alternatively, if you really know what you're doing with the stats, you could use a generalized linear mixed model, which has been shown to produce slightly more accurate rankings for differential expression because it takes into account random effects (though take this with a grain of salt, since it's mostly simulated data).
Log in to answer this question.
How about "none of the above" and you instead add a "batch" or "experiment" factor to your design.
I did experiment design,you mean no longer need to to do normalization?
No, just add a batch effect to your model. If you don't know what that is then, frankly, you should find a collaborator to do the analysis.
It sounds like they just have, for example, 4 replicates Control, 4 replicates with Drug A, and (possibly) 4 replicates with Drug B and so on...
So what would their batch variable be?
I haven't a clue what their experimental design is. Since "met-analysis" (sp) was part of the post, one presumes that there are batches involved (or OP doesn't know what a meta-analysis is).
I just wanted to ask about normalization,because of an article that I Recently read, It was emphasized that do normalization before doing meta-analysis.