about dispersion_plot and dispersion_plotLocal in DEseq2 analysis

Question

[Emily]

Hi everyone,

I am a starter in the RNAseq data analysis, I really have some questions about the DEseq analysis. After running the DEseq in R, I got several files, two of them are diffgene_results.csv and diffgene_resultsLocalFit.csv, which file should I use to get the differentially expressed genes？ And I also found that in all my cases, diffgene_resultsLocalFit.csv has more differentially expressed genes than in diffgene_results.csv. Should I based on the dispersion plot and dispersion Localplot to decide which file should I use? Or I should use the file which gives me more differentially expressed genes? Thank you very much in advance.

Answer 1

[Devon Ryan]

I'm guessing that you're using a galaxy workflow that someone else made rather than actually running DESeq (I hope you mean DESeq2) yourself.

In any case, I would normally use the parametric fit results (diffgene_results.csv) unless you received a warning during the fitting or can see in the plot that it fits poorly. Don't just blindly use the one that gives more DE results...that's a good way to waste large amounts of time and money.

[Emily]

Hi Devon,

Thank you very much for your help. Yes, I mean DEseq2. When I ran DEseq2, I do get some warning like NAs produced. But the parametric fit is not poor. The following is the parametric and local fit plot, so I should use the parametric fit right?

I appreciate your help.

Emily