CPM after ComBat-seq

Question

[Gum]

hi!

Are original CPMs modified after a batch correction with ComBat-seq?

(CPMs are re-calculated after the process)

Thank you

[ATpoint]

Please upvote and/or accept answers rather than adding titles like "closed" which have no meaning. Thank you.

Answer 1

[Kevin Blighe]

Most likely, yes. Can you please provide details of your complete workflow?

For example:

1. normalisation of raw counts with EdgeR
2. calculate CPM
3. batch correction
4. et cetera

[Gum]

Thank you for you comment.

My plan is described below:

1. RNAseq
2. STAR mapping
3. Htseq
4. calculate CPM(= original CPM)
5. ComBat-seq using raw count data
6. calculate CPM

Before running an RNA-seq, I don't have any data to estimate that.

Thank you in advance.

[Kevin Blighe]

Do you need Step 4? I think that you should be doing:

1. Run RNAseq experiment
2. STAR mapping
3. Htseq-count
4. ComBat-seq using raw count data
5. normalisation of batch-corrected raw counts
6. calculate CPM

[Gum]

Thank you very much for you kind advice. After that, I obtained RNA-seq data and see if CPM varies or not. As you mentioned above, I also confirmed that the CPM varies after ComBat-seq process. I should read the original article carefully.

Thank you for your kind help.

[Kevin Blighe]

Yes, it makes sense that the CPM values vary before and after ComBat-seq. ComBat-seq directly modifies the raw count values, meaning that any other calculation performed on these (including CPM (Counts per million)) will also be modified.