Accurate quantification in proteomics with QuantUMS

Abstract

In mass-spectrometry-based proteomics it remains challenging to ensure the accuracy of protein quantities. Here we introduce QuantUMS (quantification using an uncertainty-minimizing solution), a machine learning-based method that dynamically tunes the quantification algorithm to minimize quantitative errors. When applied to data-independent acquisition proteomics, QuantUMS increases accuracy and precision, ameliorates ratio compression bias and enhances differential expression analysis. It further reports an uncertainty measure enabling quality control of individual quantities.