Thank you for creating such a great package!

This is a more general question. Given that limpa is doing the imputation at the peptide level, when would one perform normalization and batch correction? Could batch correction and normalization be performed at the protein level after the imputation and then the differential expression run?

limpa produces expression quantifications at the protein level, together with observation-level SEs. Although it inputs peptide-level intensities, it does not need to do peptide-level imputations: the algorithm instead goes straight to protein level in one step. You can think of it as combining imputation and LFQ into one step.

The EList object produced by dpcQuant(), containing protein log-expression values, can be normalized by normalizeBetweenArrays(), but we find that normalization is not always necessary. Upstream peptide quantification tools like DIA-NN seem to already do some normalization, which sometimes seems to us to be sufficient.

Batch correction is done by including the batch effects in the design matrix when running dpcDE().

In summary, the order is:

• dpcQuant
• normalizeBetweenArrays (optional)
• Identification of surrogate variables (optional)
• dpcDE (including batch factors or surrogate variables)