Statistical quality assessment and outlier detection for liquid chromatography-mass spectrometry experiments

Abstract

Background: Quality assessment methods, that are common place in engineering and industrial production, are not widely spread in large-scale proteomics experiments. But modern technologies such as Multi-Dimensional Liquid Chromatography coupled to Mass Spectrometry (LC-MS) produce large quantities of proteomic data. These data are prone to measurement errors and reproducibility problems such that an automatic quality assessment and control become increasingly important.

LC-MS-based metabolomics

“…Schulz-Trieglaff et al proposed another outlier detection algorithm by first summarizing the characteristics of LC-MS raw data with 20 quality descriptors. Robust PCA is then used to reduce the 20 descriptors to a vector of 6 dimensions, and Mahalanobis distance is used to identify the LC-MS run outliers 37 .…”

Application of clinical assay quality control (QC) to multivariate proteomics data: A workflow exemplified by 2-DE QC

“…As per Section 5.2, we will go on to show that for gels effective QC can be carried out on samples directly on an ongoing basis, which would mirror MS-based methods without an intrinsic standard. Interestingly, in MS-based profiling Mahalanobis-based metrics are currently applied to individual runs, suggesting this is feasible [43,44]. The key element of our approach, resolving separate assignable effects, should also be possible.…”

“…In fact, in MS-profiling studies, Mahalanobis distance metrics are used already to characterise measures including test spots on sample chips for chip rejection, comparing replicate variability as against original standard runs, and for outlier identification using the peptide profile, e.g. [35,[42][43][44]. These provide effective QC solutions for drift measurement and our approach is related to these in the use of PCA and distance assessment, but not identical.…”

Generalized linear and mixed models for label-free shotgun proteomics

“…The list is used as an input into biological knowledge bases, such as FatiGo, Davids or IPA (Al-Shahrour et al 2004; Shah et al 2010; Schulz-Trieglaff et al 2009). Information about affected signaling pathways, protein interaction networks, protein localizations and functionality are used in a systems biology view of a disease state, progression, efficacy of treatment, etc.…”