Is there a need to batch correct FPKM or TPM values for within sample comparison

Question

[CTLong]

Hi all,

Does anyone have any insight or experience into whether FPKM/TPM expression should be corrected for batch effect? I have come to notice that batch correction is mostly applied to raw counts and is mainly for the purpose of between-sample comparisons or visualisation. However, since FPKM and TPM are mainly used for within sample comparisons, is there a need to batch correct the expression values for samples from different batches? I guess the bigger question is whether within-sample comparison is effected by batch effect. Thanks.

[ATpoint]

Depends on the analysis. If you do some sort of gene ratio calculation then probably yes. Batch correction does not affect all genes equally and at the same magnitude, so it changes ratios between genes even within a sample.

[CTLong]

That's a good point. So its probably safest to not batch correct for within sample comparisons I guess?

Answer 1

[jv]

My assumption is that intra-sample comparisons should not require batch effect correction.

[CTLong]

That's what I would assume, because we are not taking into account cross sample differences here. Thanks