Question: Linear regression in DESeq2
gravatar for chimeric
16 months ago by
chimeric0 wrote:

Is it possible to conduct a linear regression analysis in DESeq2, where one could get a slope, r-squared, and p-value?  Based on this post (DESEQ2 linear model of dose) it looks like I can get a slope (though it confuses me how log2FoldChange could be interpreted as a slope...clarification would be helpful here).

From my previous analysis in DESeq, I've identified which genes change over time relative to control using a LRT analysis.  Now I would like to identify the pattern of gene expression over time, irrespective of the control.  For example, did expression start low, and then end high?  I had planned to do a simple linear regression analysis on the normalized counts, but was wondering if it would be more appropriate to do this in DESeq2.  From the post above, it sounds like I would be able to tell the direction of the slope, but it is unclear to me if it would also have the information to tell the goodness of fit of the data, and whether the slope is significantly different from zero.

Thank you!


ADD COMMENTlink modified 16 months ago by Michael Love20k • written 16 months ago by chimeric0
gravatar for Michael Love
16 months ago by
Michael Love20k
United States
Michael Love20k wrote:
We have a FAQ about continuous covariates in the DESeq2 vignette. Briefly, yes, you can include them, and this is modeling the log of counts over changes in the covariates. The LFC is the change for each unit in the covariate. You can see this from the formula log(q) = X beta in the vignette and DESeq2 paper. Like any other covariate, you can obtain Wald or LRT results tables for this coefficient.
ADD COMMENTlink modified 16 months ago by Wolfgang Huber13k • written 16 months ago by Michael Love20k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 313 users visited in the last hour