Best protein imputation method when comparing knock-outs?

Question

[sebastian.gregoricchio]

Dear all, We performed a gene knock-out and we are analyzing proteomics mass-spectrometry label-free (LFQ) data. For the imputation usually I use MissForest. However, the knocked-out protein is not identified as differential (limma).

Indeed the imputation of the knock-out samples leads to the assignment of values similar to the wild-type ones. I believe that this happens when all the 4 replicates have NA values (because knock out).

Do you have any suggestion for a better imputation method that can handle this issue/case?

Answer 1

[KABILAN]

You can try the R package named lfproQC which will give you the suitable combination of normalization and imputation method for your dataset out of three methods in each. It's also available as a Shiny app at https://dabiniasri.shinyapps.io/lfproQC/.