Question: DESeq2 Fit Type Choice
gravatar for j.christopher.rounds
6 months ago by
j.christopher.rounds20 wrote:

Hey folks,

I am assessing results of gene-level differential expression testing with DESeq2, and I could use some help. I am trying to determine which fit type to use in (if I understand correctly) defining the curve to shrink dispersions/counts toward. I have 2 conditions (Drosophila wild type or null for gene X) with 3 replicates of each.

In looking at other forum posts on the subject, and at DESeq2 documentation, my understanding is that parametric fit (the default) is usually best for most RNA-Seq data, and matches well with the expected negative relationship between gene-level dispersion and mean count.

I’ve linked the dispersion and MA plots generated when using the parametric fit below. These data look similar to expected plots I see elsewhere – does this fit seem appropriate for my data? I’ve also included the dispersion and MA plots for local and mean fits for comparison.

Parametric Fit:

Local Fit:

Mean Fit:

Also, one of the forum posts I’ve read (link) describes an objective way to evaluate which fit type is best for a given dataset, by comparing the medians of: (absolute value((log(dispGeneEst) - log(dispFit)))) for each fit type. Can these values (dispGeneEst and dispFit) be outputted from the DESeq2 Galaxy tool, or would I need to download and run the software locally in R?

Thank you so much for your time!

fit type galaxy deseq2 • 262 views
ADD COMMENTlink written 6 months ago by j.christopher.rounds20
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 113 users visited in the last hour