A Peptide-Level Fully Annotated Data Set for Quantitative Evaluation of Precursor-Aware Mass Spectrometry Data Processing Algorithms

CHICKN: extraction of peptide chromatographic elution profiles from large scale mass spectrometry data by means of Wasserstein compressive hierarchical cluster analysis

Background

The clustering of data produced by liquid chromatography coupled to mass spectrometry analyses (LC-MS data) has recently gained interest to extract meaningful chemical or biological patterns. However, recent instrumental pipelines deliver data which size, dimensionality and expected number of clusters are too large to be processed by classical machine learning algorithms, so that most of the state-of-the-art relies on single pass linkage-based algorithms.

Results

We propose a clustering algorithm that solves the powerful but computationally demanding kernel k-means objective function in a scalable way. As a result, it can process LC-MS data in an acceptable time on a multicore machine. To do so, we combine three essential features: a compressive data representation, Nyström approximation and a hierarchical strategy. In addition, we propose new kernels based on optimal transport, which interprets as intuitive similarity measures between chromatographic elution profiles.

Conclusions

Our method, referred to as CHICKN, is evaluated on proteomics data produced in our lab, as well as on benchmark data coming from the literature. From a computational viewpoint, it is particularly efficient on raw LC-MS data. From a data analysis viewpoint, it provides clusters which differ from those resulting from state-of-the-art methods, while achieving similar performances. This highlights the complementarity of differently principle algorithms to extract the best from complex LC-MS data.

XFlow: An algorithm for extracting ion chromatograms

Mass spectrometry is a fundamental tool for modern proteomics. The increasing availability of mass spectrometry data paired with the increasing sensitivity and fidelity of the instruments necessitates new and more potent analytical methods. To that end, we have created and present XFlow, a feature detection algorithm for extracting ion chromatograms from MS1 LC-MS data. XFlow is a parameter-free procedurally agnostic feature detection algorithm that utilizes the latent properties of ion chromatograms to resolve them from the surrounding noise present in MS1 data. XFlow is designed to function on either profile or centroided data across different resolutions and instruments. This broad applicability lends XFlow strong utility as a one-size-fits-all method for MS1 analysis or target acquisition for MS2. XFlow is written in Java and packaged with JS-MS, an open-source mass spectrometry analysis toolkit.

A web-based system for creating, viewing, and editing precursor mass spectrometry ground truth data

Background

Mass spectrometry (MS) uses mass-to-charge ratios of measured particles to decode the identities and quantities of molecules in a sample. Interpretation of raw MS depends upon data processing algorithms that render it human-interpretable. Quantitative MS workflows are complex experimental chains and it is crucial to know the performance and bias of each data processing method as they impact accuracy, coverage, and statistical significance of the result. Creation of the ground truth necessary for quantitatively evaluating MS1-aware algorithms is difficult and tedious task, and better software for creating such datasets would facilitate more extensive evaluation and improvement of MS data processing algorithms.

Results

We present JS-MS 2.0, a software suite that provides a dependency-free, browser-based, one click, cross-platform solution for creating MS1 ground truth. The software retains the first version’s capacity for loading, viewing, and navigating MS1 data in 2- and 3-D, and adds tools for capturing, editing, saving, and viewing isotopic envelope and extracted isotopic chromatogram features. The software can also be used to view and explore the results of feature finding algorithms.

Conclusions

JS-MS 2.0 enables faster creation and inspection of MS1 ground truth data. It is publicly available with an MIT license at github.com/optimusmoose/jsms.

Quantitative Evaluation of Ion Chromatogram Extraction Algorithms

Extracted ion chromatograms (XIC) are the fundamental signal unit in mass spectrometry. There are many algorithms for analyzing raw mass spectrometry data tasked with distinguishing real isotopic signals from noise. While one or more of the available algorithms are typically chained together for end-to-end mass spectrometry analysis, analysis of each algorithm in isolation provides a specific measurement of the strengths and weaknesses of each approach. Though qualitative opinions on extraction algorithm performance abound, quantitative performance has never been publicly ascertained. Quantitative evaluation has not occurred partly due to the lack of an available quantitative ground truth MS1 data set. Using a recently published, manually extracted XICs as ground truth data, we evaluate the quality of popular XIC algorithms, including MaxQuant, MZMine2, and several methods from XCMS. The manually curated data set comprises 48 human proteins stratified over 6 abundance orders of magnitude. Signals in the sample were manually curated into XIC using a commercial tool for visually identifying XIC and isotopic envelopes. XIC algorithms were applied to the manually extracted data using a grid search of possible parameters. Performance varied greatly between different parameter settings, though nearly all algorithms with parameter settings optimized with respect to the number of true positives recovered over 10 000 XICs.

XNet: A Bayesian Approach to Extracted Ion Chromatogram Clustering for Precursor Mass Spectrometry Data

Liquid chromatography mass spectrometry is a popular technique for high throughput analysis of biological samples. Identification and quantification of molecular species via mass spectrometry output requires postexperimental computational analysis of the raw instrument output. While tandem mass spectrometry remains a primary method for identification and quantification, species-resolved precursor data provides a rich source of unexploited information. Several algorithms have been proposed to resolve raw precursor signals into species-resolved isotopic envelopes. Many methods are particularly dependent on user parameters, and because they lack a means to optimize parameters, tend to perform poorly. To this end we present XNet, a parameter-less Bayesian machine learning approach to isotopic envelope extraction through the clustering of extracted ion chromatograms. We evaluate the performance of XNet and other prevalent methods on a quantitative ground truth data set. XNet is publicly available with an Apache license.