Batch effect correction without detecting any clustering on batches

Question

[SHN]

Hi All,

I have a critical question regarding batch effect correction. I have 290 samples on 4 treatments and 41 batches. Batches here are the numbers given for each sequencing run. When I run PCA or MDS plot, I do not see any clustering based on the batch. I do not see any batch clustering even when I run PCA with a smaller number of samples.

I am aware of the technical variability in the RNA-seq. Though my question is this: Do I still need to do batch correction for my DE analysis regardless of PCA ? (due to having a great number of batches in my dataset?

Appreciate your input,

[ATpoint]

In my experience the sequencing run itself has little impact on the results. I guess you mean like the day or lane that the samples were run, or rather the library prep itself? If the facility or person who did the libraries did a good job you do not necessarily expect batch effects. If there is no evidence for it then don't correct it. You could run something like RUVseq to see whether it can identify hidden sources of variation but actually if you do not see them by eye, plus the large sample size, I guess it would be fine to proceed. Hard to tell, can you show a PCA biplot somehow colored to get the idea which samples belong to which group/batch?