Mixed-Effect Models for High-Throughput Mass Spectrometry-Based Proteomic Profiling Experiments

• Activity Number: 285
• Type: Topic Contributed
• Date/Time: Tuesday, August 3, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Timothy Clough and Ching-Yun (Veavi) Chang and Olga Vitek*+
• Companies: Purdue University and Purdue University and Purdue University
• Keywords: mass spectrometry ; proteomics ; mixed models ; biomarker discovery

Abstract

The goal of mass spectrometry-based proteomics is to identify and quantify proteins in complex biological mixtures. The technology allows a global profiling of proteins with liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) workflows, and a targeted profiling with selected reaction monitoring (SRM) workflows. The output of these experiments is a list of identified and quantified spectral features, and researchers typically apply ad-hoc rules to make protein-level conclusions. We discuss the application of mixed-effects linear models to protein quantification. For LC-MS workflows, we demonstrate using a spike-in and a clinical dataset that the models improve the sensitivity and specificity of testing, and the accuracy of patient-specific protein quantifications. We further extend the models to targeted SRM workflows, and discuss their open-source implementation.