This is a test version of Biostars. For the public version, visit https://www.biostars.org.
RNA seq batch correction with technical replicates

I am asked to process RNA seq data collected by a former labmate. There are 2 batches. Batch 1 has condition A (n=3) and B (n=3), batch 2 has condition B (technical replicates of B in batch 1) and condition C (n=3). Are there ways to remove batch effect and view PCA and heatmap? I was thinking of using Combat_seq, but it seems not specially dealing with technical replicates. I would also like to do differential expression comparison between condition C and A after adjustment, but I am not sure whether batch effect still exists between C and A. Any advice would be greatly appreciated!

batch-effect rna-seq technical-replicates

with technical replicates

Technical replicates of what kind? Sequencing/library prep? For sequencing technical replicates you can simply pool the data as long as the libraries were sequenced on the same kind of sequencer/flowcell.

Are A B C all part of the same experiment? Were the samples collected and processed at the same time?

First, samples for condition A and B were collected and sequenced. This is batch 1. Then after some days, samples for condition C were collected, and sent to sequencing together with the same RNA of condition B. This is batch 2. From the PCA plot, I can see the two batches are clearly separated, even the technical replicates of B are also clearly separated. So the batch effect is quite large. I hope that, with condition B as a bridge, I would be able to compare C and A with batch effect removed. What is the right way to do it? Thank you!

even the technical replicates of B are also clearly separated.

Still unclear if the sample B was made into to two sets of libraries (one each from batch 1 and 2). If they were, are they separated on PCA? That does not make sense ... unless a different kit was used and/or the operator that did the second set of libraries did something different and/or the sequencer used for second set was different.

Samples for condition B were made into two different sets of libraries for batch 1 and 2. The person who did the experiment has left, so I do not have much more information. On PCA, the B samples are separated by batch.

0 answers

No answers yet.

Log in to answer this question.