LOG2FCs representation

Question

[Akash D]

So, the problem is that DESeq2 takes the average of FCs for each replicate and then converts the averaged FC into Log2. That means it considers FCs. Now to represent I need error bars. For error bars, I need Standard deviation. For standard deviation I need either FC or Log2Fc values for each replicate. I found that the Log2FC formed from individual Log2FCs is not equal to Log2FC from average of the FCs. (Which is how DESeq2 does it)

Now, if I represent avg (Log2FC) with error bars, this will be similar but not the same as calculated by DESeq2. For example, a log2 value calculated from avg (FCs)= -2.30 And log2 value calculated by doing avg(log2FCs) of the same data gives = -3.94 It seems, there is a significant difference in this case. Elsewise, I have to make a graph using FCs in y axis for a data which enlists DEGs in log form.

Abbreviations:

Avg= average

Can you please suggest a correct way to represent my data?

[ATpoint]

The standard error is readily calculated for you in the results() output in column lfcSE. See ?results() as well as:

https://support.bioconductor.org/p/93559/#93561

https://support.bioconductor.org/p/65318/

https://support.bioconductor.org/p/117348/

https://bioc.ism.ac.jp/packages/2.14/bioc/vignettes/DESeq2/inst/doc/beginner.pdf

From the latter:

The column log2FoldChange is the effect size estimate. (...) Of course, this estimate has an uncertainty associated with it, which is available in the column lfcSE, the standard error estimate for the log2 fold change estimate.

[Akash D]

Loved it! Thanks.