Nonlinear Regression Improves Accuracy of Characterization of Multiplexed Mass Spectrometric Assays

Cyril Galitzine,Jarrett D Egertson,Susan Abbatiello,Clark M Henderson,Lindsay K Pino,Michael Maccoss,Andrew N Hoofnagle,Olga Vitek

doi:10.1074/mcp.ra117.000322

Abstract

The need for assay characterization is ubiquitous in quantitative mass spectrometry-based proteomics. Among many assay characteristics, the limit of blank (LOB) and limit of detection (LOD) are two particularly useful figures of merit. LOB and LOD are determined by repeatedly quantifying the observed intensities of peptides in samples with known peptide concentrations and deriving an intensity versus concentration response curve. Most commonly, a weighted linear or logistic curve is fit to the intensity-concentration response, and LOB and LOD are estimated from the fit. Here we argue that these methods inaccurately characterize assays where observed intensities level off at low concentrations, which is a common situation in multiplexed systems. This manuscript illustrates the deficiencies of these methods, and proposes an alternative approach based on nonlinear regression that overcomes these inaccuracies. We evaluated the performance of the proposed method using computer simulations and using eleven experimental data sets acquired in Data-Independent Acquisition (DIA), Parallel Reaction Monitoring (PRM), and Selected Reaction Monitoring (SRM) mode. When the intensity levels off at low concentrations, the nonlinear model changes the estimates of LOB/LOD upwards, in some data sets by 20-40%. In absence of a low concentration intensity leveling off, the estimates of LOB/LOD obtained with nonlinear statistical modeling were identical to those of weighted linear regression. We implemented the nonlinear regression approach in the open-source R-based software MSstats, and advocate its general use for characterization of mass spectrometry-based assays.

Highlights

To place the proposed approach in the context of existing work, we describe the definitions, notation, and current approaches for assay characterization based on linear [9] and logistic [21] curve fit with unequal variance
We evaluated the performance of the proposed approach using eleven experimental data sets, acquired in Selected Reaction Monitoring (SRM), Parallel Reaction Monitoring (PRM), or Data- Independent Acquisition (DIA), as well as using computer simulation
The Proposed Approach Improved the Accuracy of limit of blank (LOB)/limit of detection (LOD) Estimates—As we have seen in Fig. 4 and in Fig. 5, the differences in the model fit between the existing and the proposed approaches impacted the estimates of the figures of merit

Summary

Introduction

To place the proposed approach in the context of existing work, we describe the definitions, notation, and current approaches for assay characterization based on linear [9] and logistic [21] curve fit with unequal variance. We assume that the characterization is performed at the peptide level (i.e. after summarizing all the transitions or fragments of the peptide) and describe the model for a single representative peptide. Input Data and Notation—In mass spectrometry-based assay characterization experiments, each peptide is spiked or titrated in L samples in known and distinct concentrations C1 Ͻ . The spectral peaks corresponding to the peptide in a run are detected and quantified (e.g. by integrating the area under their chromatographic curve, or by any other method of choice) using a data processing tool such as Skyline [20]. The peak intensities are summarized (e.g. with the sum of the chromatographic peak areas of all the peptide transitions or fragments) and normalized, to produce a single intensity value per replicate run

Methods

Results

Conclusion