Question

limma/voom for protein intensity data

0

Entering edit mode

Endre Sebestyén ▴ 70

@endre-sebestyen-2707

Last seen 13 months ago

Hungary, Budapest

Hi all, I'm using limma to analyze a proteomics dataset, basically following the approach described here, so log2(count+1), quantile normalization, then a limma pipeline with eBayes(trend=TRUE, robust=TRUE). However, I was wondering if it is possible to use vooma on the data somehow, and/or include some precision weights or covariates based on peptide counts and sequence coverage. The peptide count ranges betwen 1 - 185 and the sequence coverage is betwen 0.2 - 90. I trust low peptide count and low sequence coverage less, and would like to use this information during model fitting.

limma proteomics voom • 2.5k views

ADD COMMENT • link updated 3.3 years ago by ATpoint ★ 4.6k • written 3.3 years ago by Endre Sebestyén ▴ 70

0

Entering edit mode

I think the issue you describe is pretty much addressed by https://bioconductor.org/packages/release/bioc/html/DEqMS.html

DEqMS builds on top of Limma, a widely-used R package for microarray data analysis (Smyth G. et al 2004), and improves it with proteomics data specific properties, accounting for variance dependence on the number of quantified peptides or PSMs for statistical testing of differential protein expression.

Limma assumes a common prior variance for all proteins, the function spectraCounteBayes in DEqMS package estimate prior variance for proteins quantified by different number of PSMs.

ADD REPLY • link 3.3 years ago ATpoint ★ 4.6k