How should i handle batch effects in proteomics differential expression?

[Sofia.B]

I'm trying to interpret the results from my proteomics assay, but I'm stuck on how to properly account for batch effects introduced during sample preparation. My differential expression analysis feels off because the variance seems artificially inflated between the two processing runs, even after a basic median normalization.

[Gregory.J]

Been there. After two prep runs the differential results looked jittery and the variance between runs was larger than expected. Median normalization helped a bit but the spread persisted. I tried a batch correction approach (ComBat-ish) but in proteomics that can overcorrect and erase real signals, so I dropped it after a careful check. I ended up sticking batch into the design and looking at residuals, which clarified that some proteins were still biased by run.

[Jeffrey.M]

We also inserted a few pooled QC samples across runs to gauge drift and plotted performance over time. That step made the drift pattern obvious and helped decide if corrections were safe.

[Jeffrey_S]

Did you randomize the sample order and include a couple technical replicates across runs?

[Layla.L]

Adding batch as a covariate did shave off some variance but it also muted some signals I’m pretty sure were real. It felt like a balancing act with no clean answer yet.