Hi,

we are running a dexseq analysis on some human rna-Seq data. we have 30 samples (a 10 samples per group). When I calculate the dispersion estimation I get a clear trend, which show the strong divergence of the data.

My question regards the bulk of genes/exon (Am I correct in assuming, DEXSeq plot the exons and not the genes?) at the lower level of the plot.

If I interpret the plot correctly, the exons at the bottom of the plot show a strong dispersion compared to the bulk of the data. As they are not circled blue, I assume the dispersion was "corrected".

Is there a way of telling which of the samples cause this strong divergence? 

thanks

Assa

Dear Assa

faced with such a dispersion plot, I would be worried that there be quality problems in the data (failed libraries, libraries with very small coverage, very uneven coverage, truncated files, ..., you name it). Have you done some exploratory analysis (pairwise MA plots, correlation heatmap, heatmap of VST transformed data matrix, PCA, raw data plots of some exemplary genes) to explore that?

Wolfgang