Part of the appeal of an RUV-like approach compared to something like (vanilla) SVA is the use of control genes or replicate samples to provide better estimates of technical vs. biological variability. A comparison between the 'ssva' method in the preprint using control genes with SVA and the RUV approach would be really interesting.
Does tools for removing batch effect from microarray data work for RNA-seq data as well ?
Hi all,
I want to estimate and remove batch effect from RNA-seq data. RNA-seq data is count data and has skewed distribution, because of this fact, I want to know that, whether tools for removing batch effect from microarray data are also applicable to RNA-seq data?
• 3,843 views
•
link
3 answers
If you work with something like log2(RPKM + 0.1) values and you know about the underlying causes of the batch effects, then I would use something like a normal ANOVA model to take the confounding variables into consideration.
If you don't know the confounding variables, you can try using tools like SVA. However, that may also remove true variability, so I think the effectiveness will vary between datasets.
• 0 views
•
link
• 0 views
•
link
Log in to answer this question.