Ngs Of Mirna, Differential Expression With No Replicate Sample

Question

[Sara]

Hi every one,

I am so happy to find you as I am new in NGS area. I have a set of miRNA data coming from Solid sequencer. I want to do the differential expression analysis of data. In fact, I have two groups and have 1 library per group without replicate. I want to know which statistical model fit better with my experiment Have any idea, please tell me! I have already looked at the library edger and DESeq Thanks Sara

[User 59]

I don't think a lack of replicates lends itself well to any statistical model I'm afraid.

[Jeremy Leipzig]

What was wrong with the approaches provided by DESeq and EdgeR?

[Dejian]

I am also wondering what's wrong with the model of negative binomial distribution implemented in edgeR. I once used it to analyze RNA-seq data with only one replicate for each treatment. There is a discussion on SEQanswers in which you may be interested: http://seqanswers.com/forums/showthread.php?t=3951

Answer 1

[David Quigley]

Daniel's comment is correct. With no replicates of any kind, you will not be able to find any statistical model that supports differential expression analysis. You should read up on experimental design for microarray experiments, as there is a large literature of books for beginners to microarray analysis and the underlying ideas are the same even though the specific technology is different.