Hi,
I am a newbie in R and edgeR. I am calculating DEs in different samples. For example I have 6 paired samples of drug treatment. I want to test for DEs before and after. I found 110 DEs after drug administration. Now I want to see the logFC of those 110 genes in each individual to check if they all change in the same direction. How can I do that? I have used glm approach for paired samples. Thanks in advance.
1 answer
Well, there is this option in edgeR to calculate logFC when there is no replicates. Can I run that test on each individual samples assuming there is not replicates? But this model requires an estimation of BCV. Can I use the BCV calculated from all six sample analysed together to be used in this model. edgeR manual section 2.11
Log in to answer this question.
-
z-score of gene set
written by Hien •
-
How to perform DESeq2 based on paired sample test?
written by wmsalsah •
-
Power analysis to determine sample size for a single-cell sequencing study
written by halo22
-
Does anyone explain step by step how obtain Fc from Tmm normalizatio count???
written by g.persic1991 •
-
How to get differential gene expression within paired samples using EdgeR or DESeq2?
written by learner.bioinformatics •
-
CPM plot for individual genes
written by iraia.munoa
-
Inferring Inter-individual variation using only merged raw read counts of the individual genes ?
written by sandKings •
-
Time-series or pairwise comparison EdgeR - what's more suited?
written by Biogeek
-
EdgeR filter thresholds
written by Biogeek
-
Gene-Level Analysis Of Rna-Seq Matched Pairs Of Samples?
written by Ryan Thompson
On what basis you are saying 110 genes are deferentially expressed ? I think if you know the answer, you can filter the DE object based on fold change and significance level. pardon me If i am missing something.
It would only really be possible to do this if you rewrote your design with a different factor level for each of your 12 samples, and did a within-patient contrast for each patient. I suspect edgeR would complain due to the lack of degrees-of-freedom in the model, howver, I believe other people do use edgeR without replicates so it maybe possilbe.
I don't think it adds much to your analysis though, and will probably lead to you throwing out some really good candidates.
thank you very much. No I just want to see if the fold change in the DE matches in each individual as I assume the fold change in the final DE list may be an average of all 6 individuals. Will you please explain the design matrix a little bit more?
Thank you very much for your reply. I have run the glm model from edgeR following the example given in the manual section 4.1. using patient and treatment as additive factors. Now I want to see the foldchange of those genes in each individual.
You can't see logFC in each individual, because logFC are ratios - you need at least two individuals for the comparison. Maybe you want the counts for each gene for each individual? You can use edgeR function
cpm().I think the logFC alluded to is drug vs no-drug in a given individual
yes, that is what I want to check. Any suggestions how to do that?
logFC values in edgeR are not reliable, if you want to include FC anyway in your edgeR analysis, you'll have to use
treatfunction. For detailed questions about why logFCs are not reliable in edgeR please ask them at bioconductor, since the authors of the package only answer the question there, not here at biostars.