This is a test version of Biostars. For the public version, visit https://www.biostars.org.
How to detect and correct batch effect in NGS expression data ?

Hi all,

I have RNA-Seq gene expression data for a cancer. It's around 1000 samples. I think it's been in 50 different batches.

I have two questions:

  1. How can I check/test that, whether there is any batch effect on my data
  2. How can I correct it?
next-gen sequencing r rna-seq

2 answers

  1. Have a look at a PCA plot.
  2. Give the SVA package from Bioconductor a try.

SVA has just been updated, apparently, for use with sequencing data; see this preprint.

The ssva functionality looks particularly interesting.

It sort of depends on your goals and circumstances. If you have known batch effects that and can use an expression-set as output, then combat would seem the better choice. If you need to known how many unknown batch effects you might have and appropriate weights for the various samples then sva is a better choice.

edX course HarvardX: PH525x Data Analysis for Genomics week6 for removing batch effect https://courses.edx.org/courses/HarvardX/PH525x/1T2014/courseware/2273065cc0f649b69c1240a58f7ab080/

Log in to answer this question.