Batch Effect In Rnaseq

Question

[kanwarjag]

I am working on very small no of tumor cells coming from patients and obviously cannot process all at one time. SO may be 2-3 batches. How can I correct fro a batch effect? How I will know that there is some batch effect that may effect my out come?

Thanks

Answer 1

[Mikael Huss]

You can check for batch effects by making a PCA or MDS plot, or a correlation heat map, from normalized count or FPKM values. This could be done in DESeq, edgeR or just "manual" R code. If you see samples cluster by batch in these plots, you probably have a batch effect (unless there is some other plausible reason for them to cluster that way).

One way to attempt to correct for batch effect is to include the batch as a factor in the design matrix in a multi-factorial analysis in DESeq or edgeR. The user's guides should tell you how to do that. Obviously this will not remove all batch effects but it may help.