Normalization of RNA seq expression values between different experiments

Question

[qstefano]

Hello there,

I have different E.Coli RNA-seq experiments data, i need to compare them to find which genes are not differentially expressed. In each experiment there are several conditions, each condition have several replicates. First i used DESeq normalization for gene expression values between conditions, so i get normalized values for every experiments. Now i need to do the same thing between experiments (the experiments come from the same organism, but may change for sequencing technology).

The question is: there's a method which can perform that? Can i eventually reuse DESeq without introducing bias?

[ATpoint]

The general and simplified rule is that you need samples of every treatment group in every batch (that is here study you downloaded it from) to perform batch correction (that is the term that is of interest here). Is that the case? Can you post a table indicating which sample belongs to which study and protocol type?

Answer 1

[Qiongyi]

In general, edgeR or DESeq2 should be fine. You need PCA analysis to check if there are any obvious outliers due to different sequencing technologies. I will be very concerned if all samples in group A were sequenced using one sequencing technology and samples in group B were sequenced using another technology, and if you want to compare group A with group B...