Search
Question: Estimation of dispersion in DeSeq and DeSeq2
1
gravatar for JK
4.1 years ago by
JK10
United States
JK10 wrote:

I have a couple of questions about estimation of dispersion in DeSeq2 vs. DeSeq. The paper and manual on DeSeq and DeSeq2 are very clear but I have a hard time understanding what is the ‘correct’ or ‘recommended’ method of estimating the dispersion is for RNA-Seq. Does this depend on the sample size? Are there other assumptions that have to be met for each dispersion estimate type (mean, local or parametric)? is there a particular reason behind ‘tag-wise’ estimation not being in DeSeq2?

Thank you!

ADD COMMENTlink modified 4.1 years ago by Michael Love20k • written 4.1 years ago by JK10
1
gravatar for Michael Love
4.1 years ago by
Michael Love20k
United States
Michael Love20k wrote:

Hi JK,

For DESeq2, we have 3 options for the type of trend line: fitType = "parametric", "local" or "mean". These are described in ?estimateDispersions, but roughly the recommendation is: for decreasing gene-wise dispersion estimates over mean (using plotDispEsts) one should use parametric, unless the parametric fitting procedure does not work, in which case use "local" (local regression is actually automatically substituted with a message in the case that the parametric fitting procedure does not converge.) The "mean" option is useful when there is no apparent dependence of dispersion estimates over mean (using plotDispEsts). This choice does not depend on sample size, but on the apparent dependence of the gene-wise estimates (the MLE for each gene) on the mean of counts. 

If you are referring to the the tagwise estimation in edgeR, the tagwise estimation is similar to the estimateDispersionsMAP() step in DESeq2, which is the last step in estimateDispersions(), which is automatically used by DESeq().

ADD COMMENTlink written 4.1 years ago by Michael Love20k

Michael,

Thank you for your answer. This clears it up for me.

Best,

JK

ADD REPLYlink written 4.1 years ago by JK10

Michael,

Is there a reason that the per-sample dispersion calculation was discontinued for DESeq2? I am working with ASD samples which have been observed [ http://dx.doi.org/10.1371/journal.pone.0016715 ] to exhibit lower variance. I need to estimate the ASD and typical dispersion separately. Can I do this with DESeq2?

Ben

ADD REPLYlink written 2.1 years ago by bkellman0

It was replaced because we wanted to generalize all the methods for arbitrary experimental designs. DESeq2 doesn't have support for each condition having it's own dispersion. However, you can use DESeq2 to estimate the dispersion of a set of samples, if that is what you are interested in comparing. Create a dds object with the samples of a single condition and use a design of ~1, then run estimateSizeFactors() and estimateDispersions() and you can access the dispersions with the dispersions() function.

ADD REPLYlink written 2.1 years ago by Michael Love20k

Thanks! Let me repost this. I think this is a useful exchange and I need some clarification

ADD REPLYlink written 2.1 years ago by bkellman0

Thanks! Let me repost this. I think this is a useful exchange and I need some clarification

Estimating group-specific dispersion in DESeq2

ADD REPLYlink written 2.1 years ago by bkellman0
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 339 users visited in the last hour