Hi everyone,
Apologies if this has been asked before and for the length of this post, but there seems to be a variety of answers out there for different types of studies, so I wasn't sure if my approach was correct.
I have 3 RNA-seq studies where I have performed DE using DESeq2. I want to perform a meta-analysis using the metafor R package and then create a forest plot along with a summary variable.
My main question is what do I use as the effect size and standard error when I am performing a meta-analysis? Currently I am using the values obtained from DESeq2. Is this correct to do? So for example:
#Obtain the logFC and lfcse from each dataset for gene of interest.
Study LogFC lfcSE
StudyA -1.27 0.22
StudyB -1.17 0.24
StudyC 1.87 0.37
#Calculate Effect Size for Gene of Interest:
StudyA <- ALogFC / AlfcSE
StudyB <- BLogFC / BlfcSE
StudyC <- CLogFC / ClfcSE
Now I am a little confused with calculating the standard error. At the moment I am calculating it through the following:
SE <- sqrt((AlfcSE^2 + BlfcSE^2 + Clfc3^2) / k) #where k = # of studies (3)
Then I would also have to calculate the Upper and Lower CI values for the forest plot, which would be: Upper value = meta_effect + z-value x meta_se Lower value = meta_effect - z-value x meta_se
where:
meta_effect: the estimated meta-analytic effect size, which is the weighted average of the effect sizes from the individual studies meta_se: the estimated standard error of the meta-analytic effect size z-value: the critical value from the standard normal distribution for the desired level of significance (e.g., z = 1.96 for a 95% confidence interval) For example, if the meta-analytic effect size is 1.5, the estimated standard error is 0.3, and the desired level of significance is 95%, then the upper and lower values would be:
Upper value = 1.5 + 1.96 x 0.3 = 2.08 Lower value = 1.5 - 1.96 x 0.3 = 0.92
Once I obtain the correct values, I would be running the "rma" function in the metafor package to obtain the summary. The "upper" and "lower" values are for the forest plot.
But I am not sure if I am doing this correctly.
Thanks in advance for all the help and for reading!
