A review of statistical methods for protein identification using tandem mass spectrometry

Acquisition and Analysis of DIA-Based Proteomic Data: A Comprehensive Survey in 2023

Data-independent acquisition (DIA) mass spectrometry (MS) has emerged as a powerful technology for high-throughput, accurate, and reproducible quantitative proteomics. This review provides a comprehensive overview of recent advances in both the experimental and computational methods for DIA proteomics, from data acquisition schemes to analysis strategies and software tools. DIA acquisition schemes are categorized based on the design of precursor isolation windows, highlighting wide-window, overlapping-window, narrow-window, scanning quadrupole-based, and parallel accumulation-serial fragmentation-enhanced DIA methods. For DIA data analysis, major strategies are classified into spectrum reconstruction, sequence-based search, library-based search, de novo sequencing, and sequencing-independent approaches. A wide array of software tools implementing these strategies are reviewed, with details on their overall workflows and scoring approaches at different steps. The generation and optimization of spectral libraries, which are critical resources for DIA analysis, are also discussed. Publicly available benchmark datasets covering global proteomics and phosphoproteomics are summarized to facilitate performance evaluation of various software tools and analysis workflows. Continued advances and synergistic developments of versatile components in DIA workflows are expected to further enhance the power of DIA-based proteomics.

Overview and considerations in bottom-up proteomics

Proteins are the key biological actors within cells, driving many biological processes integral to both healthy and diseased states. Understanding the depth of complexity represented within the proteome is crucial to our scientific understanding of cellular biology and to provide disease specific insights for clinical applications. Mass spectrometry-based proteomics is the premier method for proteome analysis, with the ability to both identify and quantify proteins. Although proteomics continues to grow as a robust field of bioanalytical chemistry, advances are still necessary to enable a more comprehensive view of the proteome. In this review, we provide a broad overview of mass spectrometry-based proteomics in general, and highlight four developing areas of bottom-up proteomics: (1) protein inference, (2) alternative proteases, (3) sample-specific databases and (4) post-translational modification discovery.

Montane Rattlesnakes in México: Venoms of Crotalus tancitarensis and Related Species within the Crotalus intermedius Group

The Crotalus intermedius group is a clade of rattlesnakes consisting of several species adapted to a high elevation habitat, primarily in México. Crotalus tancitarensis was previously classified as C. intermedius, until individuals occurring on Cerro Tancítaro in Michoacán, México, were reevaluated and classified as a new species (C. tancitarensis) based on scale pattern and geographic location. This study aimed to characterize the venom of C. tancitarensis and compare the venom profile to those of other species within the Crotalus intermedius group using gel electrophoresis, biochemical assays, reverse-phase high performance liquid chromatography, mass spectrometry, and lethal toxicity (LD50) assays. Results show that the venom profiles of species within the Crotalus intermedius group are similar, but with distinct differences in phospholipase A2 (PLA2), metalloproteinase PI (SVMP PI), and kallikrein-like serine proteinase (SVSP) activity and relative abundance. Proteomic analysis indicated that the highland forms produce venoms with 50-60 protein isoforms and a composition typical of type I rattlesnake venoms (abundant SVMPs, lack of presynaptic PLA2-based neurotoxins), as well as a diversity of typical Crotalus venom components such as serine proteinases, PLA2s, C-type lectins, and less abundant toxins (LAAOs, CRiSPs, etc.). The overall venom profile of C. tancitarensis appears most similar to C. transversus, which is consistent with a previous mitochondrial DNA analysis of the Crotalus intermedius group. These rattlesnakes of the Mexican highlands represent a radiation of high elevation specialists, and in spite of divergence of species in these Sky Island habitats, venom composition of species analyzed here has remained relatively conserved. The majority of protein family isoforms are conserved in all members of the clade, and as seen in other more broadly distributed rattlesnake species, differences in their venoms are largely due to relative concentrations of specific components.

The TriMet_DB: A Manually Curated Database of the Metabolic Proteins of Triticum aestivum

Mass-spectrometry-based wheat proteomics is challenging because the current interpretation of mass spectrometry data relies on public databases that are not exhaustive (UniProtKB/Swiss-Prot) or contain many redundant and poor or un-annotated entries (UniProtKB/TrEMBL). Here, we report the development of a manually curated database of the metabolic proteins of Triticum aestivum (hexaploid wheat), named TriMet_DB (Triticum aestivum Metabolic Proteins DataBase). The manually curated TriMet_DB was generated in FASTA format so that it can be read directly by programs used to interpret the mass spectrometry data. Furthermore, the complete list of entries included in the TriMet_DB is reported in a freely available resource, which includes for each protein the description, the gene code, the protein family, and the allergen name (if any). To evaluate its performance, the TriMet_DB was used to interpret the MS data acquired on the metabolic protein fraction extracted from the cultivar MEC of Triticum aestivum. Data are available via ProteomeXchange with identifier PXD037709.

Reanalysis of ProteomicsDB Using an Accurate, Sensitive, and Scalable False Discovery Rate Estimation Approach for Protein Groups

Estimating false discovery rates (FDRs) of protein identification continues to be an important topic in mass spectrometry-based proteomics, particularly when analyzing very large datasets. One performant method for this purpose is the Picked Protein FDR approach which is based on a target-decoy competition strategy on the protein level that ensures that FDRs scale to large datasets. Here, we present an extension to this method that can also deal with protein groups, that is, proteins that share common peptides such as protein isoforms of the same gene. To obtain well-calibrated FDR estimates that preserve protein identification sensitivity, we introduce two novel ideas. First, the picked group target-decoy and second, the rescued subset grouping strategies. Using entrapment searches and simulated data for validation, we demonstrate that the new Picked Protein Group FDR method produces accurate protein group-level FDR estimates regardless of the size of the data set. The validation analysis also uncovered that applying the commonly used Occam's razor principle leads to anticonservative FDR estimates for large datasets. This is not the case for the Picked Protein Group FDR method. Reanalysis of deep proteomes of 29 human tissues showed that the new method identified up to 4% more protein groups than MaxQuant. Applying the method to the reanalysis of the entire human section of ProteomicsDB led to the identification of 18,000 protein groups at 1% protein group-level FDR. The analysis also showed that about 1250 genes were represented by ≥2 identified protein groups. To make the method accessible to the proteomics community, we provide a software tool including a graphical user interface that enables merging results from multiple MaxQuant searches into a single list of identified and quantified protein groups.

Enhanced protein isoform characterization through long-read proteogenomics

BACKGROUND

The detection of physiologically relevant protein isoforms encoded by the human genome is critical to biomedicine. Mass spectrometry (MS)-based proteomics is the preeminent method for protein detection, but isoform-resolved proteomic analysis relies on accurate reference databases that match the sample; neither a subset nor a superset database is ideal. Long-read RNA sequencing (e.g., PacBio or Oxford Nanopore) provides full-length transcripts which can be used to predict full-length protein isoforms.

RESULTS

We describe here a long-read proteogenomics approach for integrating sample-matched long-read RNA-seq and MS-based proteomics data to enhance isoform characterization. We introduce a classification scheme for protein isoforms, discover novel protein isoforms, and present the first protein inference algorithm for the direct incorporation of long-read transcriptome data to enable detection of protein isoforms previously intractable to MS-based detection. We have released an open-source Nextflow pipeline that integrates long-read sequencing in a proteomic workflow for isoform-resolved analysis.

CONCLUSIONS

Our work suggests that the incorporation of long-read sequencing and proteomic data can facilitate improved characterization of human protein isoform diversity. Our first-generation pipeline provides a strong foundation for future development of long-read proteogenomics and its adoption for both basic and translational research.

Novel Bioinformatics Strategies Driving Dynamic Metaproteomic Studies

Constant improvements in mass spectrometry technologies and laboratory workflows have enabled the proteomics investigation of biological samples of growing complexity. Microbiomes represent such complex samples for which metaproteomics analyses are becoming increasingly popular. Metaproteomics experimental procedures create large amounts of data from which biologically relevant signal must be efficiently extracted to draw meaningful conclusions. Such a data processing requires appropriate bioinformatics tools specifically developed for, or capable of handling metaproteomics data. In this chapter, we outline current and novel tools that can perform the most commonly used steps in the analysis of cutting-edge metaproteomics data, such as peptide and protein identification and quantification, as well as data normalization, imputation, mining, and visualization. We also provide details about the experimental setups in which these tools should be used.

Critical Assessment of MetaProteome Investigation (CAMPI): a multi-laboratory comparison of established workflows

Metaproteomics has matured into a powerful tool to assess functional interactions in microbial communities. While many metaproteomic workflows are available, the impact of method choice on results remains unclear. Here, we carry out a community-driven, multi-laboratory comparison in metaproteomics: the critical assessment of metaproteome investigation study (CAMPI). Based on well-established workflows, we evaluate the effect of sample preparation, mass spectrometry, and bioinformatic analysis using two samples: a simplified, laboratory-assembled human intestinal model and a human fecal sample. We observe that variability at the peptide level is predominantly due to sample processing workflows, with a smaller contribution of bioinformatic pipelines. These peptide-level differences largely disappear at the protein group level. While differences are observed for predicted community composition, similar functional profiles are obtained across workflows. CAMPI demonstrates the robustness of present-day metaproteomics research, serves as a template for multi-laboratory studies in metaproteomics, and provides publicly available data sets for benchmarking future developments.

Critical Assessment of MetaProteome Investigation (CAMPI): a multi-laboratory comparison of established workflows

Metaproteomics has matured into a powerful tool to assess functional interactions in microbial communities. While many metaproteomic workflows are available, the impact of method choice on results remains unclear. Here, we carry out a community-driven, multi-laboratory comparison in metaproteomics: the critical assessment of metaproteome investigation study (CAMPI). Based on well-established workflows, we evaluate the effect of sample preparation, mass spectrometry, and bioinformatic analysis using two samples: a simplified, laboratory-assembled human intestinal model and a human fecal sample. We observe that variability at the peptide level is predominantly due to sample processing workflows, with a smaller contribution of bioinformatic pipelines. These peptide-level differences largely disappear at the protein group level. While differences are observed for predicted community composition, similar functional profiles are obtained across workflows. CAMPI demonstrates the robustness of present-day metaproteomics research, serves as a template for multi-laboratory studies in metaproteomics, and provides publicly available data sets for benchmarking future developments.

Data-Independent Acquisition Mass Spectrometry of the Human Lens Enhances Spatiotemporal Measurement of Fiber Cell Aging

The ocular lens proteome undergoes post-translational and progressive degradation as fiber cells age. The oldest fiber cells and the proteins therein are present at birth and are retained through death. Transparency of the lens is maintained in part by the high abundance Crystallin family proteins (up to 300 mg/mL), which establishes a high dynamic range of protein abundance. As a result, previous data-dependent analysis (DDA) measurements of the lens proteome are less equipped to identify the lowest abundance proteins. To probe more deeply into the lens proteome, we measured the insoluble lens proteome of an 18-year-old human with DDA and data-independent analysis (DIA) methods. By applying more recent library-free DIA search methods, 5,161 protein groups, 50,386 peptides, and 4,960 deamidation sites were detected: significantly outperforming the quantity of identifications in using DDA and pan-human DIA library searches. Finally, by segmenting the lens into multiple fiber cell-age-related regions, we uncovered cell-age-related changes in proteome composition and putative function.

Heterogeneity in proline hydroxylation of fibrillar collagens observed by mass spectrometry

Collagen is the major protein in the extracellular matrix and plays vital roles in tissue development and function. Collagen is also one of the most processed proteins in its biosynthesis. The most prominent post-translational modification (PTM) of collagen is the hydroxylation of Pro residues in the Y-position of the characteristic (Gly-Xaa-Yaa) repeating amino acid sequence of a collagen triple helix. Recent studies using mass spectrometry (MS) and tandem MS sequencing (MS/MS) have revealed unexpected hydroxylation of Pro residues in the X-positions (X-Hyp). The newly identified X-Hyp residues appear to be highly heterogeneous in location and percent occupancy. In order to understand the dynamic nature of the new X-Hyps and their potential impact on applications of MS and MS/MS for collagen research, we sampled four different collagen samples using standard MS and MS/MS techniques. We found considerable variations in the degree of PTMs of the same collagen from different organisms and/or tissues. The rat tail tendon type I collagen is particularly variable in terms of both over-hydroxylation of Pro in the X-position and under-hydroxylation of Pro in the Y-position. In contrast, only a few unexpected PTMs in collagens type I and type III from human placenta were observed. Some observations are not reproducible between different sequencing efforts of the same sample, presumably due to a low population and/or the unpredictable nature of the ionization process. Additionally, despite the heterogeneous preparation and sourcing, collagen samples from commercial sources do not show elevated variations in PTMs compared to samples prepared from a single tissue and/or organism. These findings will contribute to the growing body of information regarding the PTMs of collagen by MS technology, and culminate to a more comprehensive understanding of the extent and the functional roles of the PTMs of collagen.

Quantitative Mass Spectrometry-Based Proteomics: An Overview

In recent decades, mass spectrometry has moved more than ever before into the front line of protein-centered research. After being established at the qualitative level, the more challenging question of quantification of proteins and peptides using mass spectrometry has become a focus for further development. In this chapter, we discuss and review actual strategies and problems of the methods for the quantitative analysis of peptides, proteins, and finally proteomes by mass spectrometry. The common themes, the differences, and the potential pitfalls of the main approaches are presented in order to provide a survey of the emerging field of quantitative, mass spectrometry-based proteomics.

New mixture models for decoy-free false discovery rate estimation in mass spectrometry proteomics

MOTIVATION

Accurate estimation of false discovery rate (FDR) of spectral identification is a central problem in mass spectrometry-based proteomics. Over the past two decades, target-decoy approaches (TDAs) and decoy-free approaches (DFAs) have been widely used to estimate FDR. TDAs use a database of decoy species to faithfully model score distributions of incorrect peptide-spectrum matches (PSMs). DFAs, on the other hand, fit two-component mixture models to learn the parameters of correct and incorrect PSM score distributions. While conceptually straightforward, both approaches lead to problems in practice, particularly in experiments that push instrumentation to the limit and generate low fragmentation-efficiency and low signal-to-noise-ratio spectra.

RESULTS

We introduce a new decoy-free framework for FDR estimation that generalizes present DFAs while exploiting more search data in a manner similar to TDAs. Our approach relies on multi-component mixtures, in which score distributions corresponding to the correct PSMs, best incorrect PSMs and second-best incorrect PSMs are modeled by the skew normal family. We derive EM algorithms to estimate parameters of these distributions from the scores of best and second-best PSMs associated with each experimental spectrum. We evaluate our models on multiple proteomics datasets and a HeLa cell digest case study consisting of more than a million spectra in total. We provide evidence of improved performance over existing DFAs and improved stability and speed over TDAs without any performance degradation. We propose that the new strategy has the potential to extend beyond peptide identification and reduce the need for TDA on all analytical platforms.

AVAILABILITYAND IMPLEMENTATION

https://github.com/shawn-peng/FDR-estimation.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Beyond Target-Decoy Competition: Stable Validation of Peptide and Protein Identifications in Mass Spectrometry-Based Discovery Proteomics

In bottom-up discovery proteomics, target-decoy competition (TDC) is the most popular method for false discovery rate (FDR) control. Despite unquestionable statistical foundations, this method has drawbacks, including its hitherto unknown intrinsic lack of stability vis-à-vis practical conditions of application. Although some consequences of this instability have already been empirically described, they may have been misinterpreted. This article provides evidence that TDC has become less reliable as the accuracy of modern mass spectrometers improved. We therefore propose to replace TDC by a totally different method to control the FDR at the spectrum, peptide, and protein levels, while benefiting from the theoretical guarantees of the Benjamini-Hochberg framework. As this method is simpler to use, faster to compute, and more stable than TDC, we argue that it is better adapted to the standardization and throughput constraints of current proteomic platforms.

Review: Detection and quantification of proteins in human urine

Extensive medical research showed that patients, with high protein concentration in urine, have various kinds of kidney diseases, referred to as proteinuria. Urinary protein biomarkers are useful for diagnosis of many health conditions – kidney and cardio vascular diseases, cancers, diabetes, infections. This review focuses on the instrumental quantification (electrophoresis, chromatography, immunoassays, mass spectrometry, fluorescence spectroscopy, the infrared spectroscopy, and Raman spectroscopy) of proteins (the most of all albumin) in human urine matrix. Different techniques provide unique information on what constituents of the urine are. Due to complex nature of urine, a separation step by electrophoresis or chromatography are often used for proteomics study of urine. Mass spectrometry is a powerful tool for the discovery and the analysis of biomarkers in urine, however, costs of the analysis are high, especially for quantitative analysis. Immunoassays, which often come with fluorescence detection, are major qualitative and quantitative tools in clinical analysis. While Infrared and Raman spectroscopies do not give extensive information about urine, they could become important tools for the routine clinical diagnostics of kidney problems, due to rapidness and low-cost. Thus, it is important to review all the applicable techniques and methods related to urine analysis. In this review, a brief overview of each technique's principle is introduced. Where applicable, research papers about protein determination in urine are summarized with the main figures of merits, such as the limit of detection, the detectable range, recovery and accuracy, when available.

•

Urinary protein biomarkers are useful for diagnosis of many conditions: kidney and cardio vascular diseases, cancers.

•

Liquid chromatography – mass spectroscopy is a powerful tool for urine proteomics, but used mostly in science.

•

Immunoassays are widely used in both clinical and bio-analytical laboratories.

•

IR and Raman spectroscopies are promising tools for diagnostics of urine due to low-cost and rapidness.

pepFunk: a tool for peptide-centric functional analysis of metaproteomic human gut microbiome studies

Motivation

Enzymatic digestion of proteins before mass spectrometry analysis is a key process in metaproteomic workflows. Canonical metaproteomic data processing pipelines typically involve matching spectra produced by the mass spectrometer to a theoretical spectra database, followed by matching the identified peptides back to parent-proteins. However, the nature of enzymatic digestion produces peptides that can be found in multiple proteins due to conservation or chance, presenting difficulties with protein and functional assignment.

Results

To combat this challenge, we developed pepFunk, a peptide-centric metaproteomic workflow focused on the analysis of human gut microbiome samples. Our workflow includes a curated peptide database annotated with Kyoto Encyclopedia of Genes and Genomes (KEGG) terms and a gene set variation analysis-inspired pathway enrichment adapted for peptide-level data. Analysis using our peptide-centric workflow is fast and highly correlated to a protein-centric analysis, and can identify more enriched KEGG pathways than analysis using protein-level data. Our workflow is open source and available as a web application or source code to be run locally.

Availability and implementation

pepFunk is available online as a web application at https://shiny.imetalab.ca/pepFunk/ with open-source code available from https://github.com/northomics/pepFunk.

Contact

dfigeys@uottawa.ca

Supplementary information

Supplementary data are available at Bioinformatics online.

Protein Probability Model for High-Throughput Protein Identification by Mass Spectrometry-Based Proteomics

Shotgun proteomics is the method of choice for high-throughput protein identification; however, robust statistical methods are essential to automatize this task while minimizing the number of false identifications. The standard method for estimating the false discovery rate (FDR) of individual identifications and keeping it below a threshold (typically 1%) is the target-decoy approach. However, numerous works have shown that FDR at the protein level may become much larger than FDR at the peptide level. The development of an appropriate scoring model to identify proteins from their peptides using high-throughput shotgun proteomics is highly needed. In this study, we present a novel protein-level scoring algorithm that uses the scores of the identified peptides and maintains all of the properties expected for a true protein probability. We also present a refinement of the picked method to calculate FDR at the protein level. These algorithms can be used together as a robust identification workflow suitable for large-scale proteomics, and we show that the identification performance of this workflow is superior to that of other widely used methods in several samples and using different search engines. Our protein probability model offers the scientific community an algorithm that is easy to integrate into protein identification workflows for the automated analysis of shotgun proteomics data.

EPIFANY: A Method for Efficient High-Confidence Protein Inference

Accurate protein inference in the presence of shared peptides is still one of the key problems in bottom-up proteomics. Most protein inference tools employing simple heuristic inference strategies are efficient but exhibit reduced accuracy. More advanced probabilistic methods often exhibit better inference quality but tend to be too slow for large data sets. Here, we present a novel protein inference method, EPIFANY, combining a loopy belief propagation algorithm with convolution trees for efficient processing of Bayesian networks. We demonstrate that EPIFANY combines the reliable protein inference of Bayesian methods with significantly shorter runtimes. On the 2016 iPRG protein inference benchmark data, EPIFANY is the only tested method that finds all true-positive proteins at a 5% protein false discovery rate (FDR) without strict prefiltering on the peptide-spectrum match (PSM) level, yielding an increase in identification performance (+10% in the number of true positives and +14% in partial AUC) compared to previous approaches. Even very large data sets with hundreds of thousands of spectra (which are intractable with other Bayesian and some non-Bayesian tools) can be processed with EPIFANY within minutes. The increased inference quality including shared peptides results in better protein inference results and thus increased robustness of the biological hypotheses generated. EPIFANY is available as open-source software for all major platforms at https://OpenMS.de/epifany.

Comparing Peptide Spectra Matches Across Search Engines

Mass spectrometry is extremely efficient for sequencing small peptides generated by, for example, a trypsin digestion of a complex mixture. Current instruments have the capacity to generate 50-100 K MSMS spectra from a single run. Of these ~30-50% is typically assigned to peptide matches on a 1% FDR threshold. The remaining spectra need more research to explain. We address here whether the 30-50% matched spectra provide consensus matches when using different database-dependent search pipelines. Although the majority of the spectra peptide assignments concur across search engines, our conclusion is that database-dependent search engines still require improvements.

Inferring Protein-Protein Interaction Networks From Mass Spectrometry-Based Proteomic Approaches: A Mini-Review

Studying protein-protein interaction networks provide key evidence for the underlying molecular mechanisms. Mass spectrometry-based proteomic approaches have been playing a pivotal role in deciphering these interaction networks, along with precise quantification for individual interactions. In this mini-review we discuss the available techniques and methods for qualitative and quantitative elucidation of protein-protein interaction networks. We then summarize the down-stream computational strategies for identification and quantification of interactions from those techniques. Finally, we highlight the challenges and limitations of current computational pipelines in eliminating false positive interactors, followed by a summary of the innovative algorithms to address these issues, along with the scope for future improvements.

Challenges and promise at the interface of metaproteomics and genomics: an overview of recent progress in metaproteogenomic data analysis

INTRODUCTION

The study of microbial communities based on the combined analysis of genomic and proteomic data - called metaproteogenomics - has gained increased research attention in recent years. This relatively young field aims to elucidate the functional and taxonomic interplay of proteins in microbiomes and its implications on human health and the environment. Areas covered: This article reviews bioinformatics methods and software tools dedicated to the analysis of data from metaproteomics and metaproteogenomics experiments. In particular, it focuses on the creation of tailored protein sequence databases, on the optimal use of database search algorithms including methods of error rate estimation, and finally on taxonomic and functional annotation of peptide and protein identifications. Expert opinion: Recently, various promising strategies and software tools have been proposed for handling typical data analysis issues in metaproteomics. However, severe challenges remain that are highlighted and discussed in this article; these include: (i) robust false-positive assessment of peptide and protein identifications, (ii) complex protein inference against a background of highly redundant data, (iii) taxonomic and functional post-processing of identification data, and finally, (iv) the assessment and provision of metrics and tools for quantitative analysis.

TagGraph reveals vast protein modification landscapes from large tandem mass spectrometry datasets

Although mass spectrometry is well suited to identifying thousands of potential protein post-translational modifications (PTMs), it has historically been biased towards just a few. To measure the entire set of PTMs across diverse proteomes, software must overcome the dual challenges of covering enormous search spaces and distinguishing correct from incorrect spectrum interpretations. Here, we describe TagGraph, a computational tool that overcomes both challenges with an unrestricted string-based search method that is as much as 350-fold faster than existing approaches, and a probabilistic validation model that we optimized for PTM assignments. We applied TagGraph to a published human proteomic dataset of 25 million mass spectra and tripled confident spectrum identifications compared to its original analysis. We identified thousands of modification types on almost 1 million sites in the proteome. We show alternative contexts for highly abundant yet understudied PTMs such as proline hydroxylation, and its unexpected association with cancer mutations. By enabling broad characterization of PTMs, TagGraph informs as to how their functions and regulation intersect.

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

Modern label-free quantitative mass spectrometry workflows are complex experimental chains for devising the composition of biological samples. With benchtop and in silico experimental steps that each have a significant effect on the accuracy, coverage, and statistical significance of the study result, it is crucial to understand the efficacy and biases of each protocol decision. Although many studies have been conducted on wet lab experimental protocols, postacquisition data processing methods have not been adequately evaluated in large part due to a lack of available ground truth data. In this study, we provide a novel ground truth data set for mass spectrometry data analysis at the precursor (MS1) signal level comprised of isolated peptide signals from UPS2, a popular complex standard for proteomics analysis, requiring more than 1000 h of manual curation. The data set consists of more than 62 million points with 1,294,008 grouped into 57,518 extracted ion chromatograms and those grouped into 14,111 isotopic envelopes. This data set can be used to evaluate many aspects of mass spectrometry data processing, including precursor mapping and signal extraction algorithms.

A quantitative and temporal map of proteostasis during heat shock in Saccharomyces cerevisiae

Temperature fluctuation is a common environmental stress that elicits a molecular response in order to maintain intracellular protein levels. Here, for the first time, we report a comprehensive temporal and quantitative study of the proteome during a 240 minute heat stress, using label-free mass spectrometry. We report temporal expression changes of the hallmark heat stress proteins, including many molecular chaperones, tightly coupled to their protein clients. A notable lag of 30 to 120 minutes was evident between transcriptome and proteome levels for differentially expressed genes. This targeted molecular response buffers the global proteome; fewer than 15% of proteins display significant abundance change. Additionally, a parallel study in a Hsp70 chaperone mutant (ssb1Δ) demonstrated a significantly attenuated response, at odds with the modest phenotypic effects that are observed on growth rate. We cast the global changes in temporal protein expression into protein interaction and functional networks, to afford a unique, time-resolved and quantitative description of the heat shock response in an important model organism.

The need for statistical contributions to bioinformatics at scale, with illustration to mass spectrometry

In their article, Morris and Baladandayuthapani clearly evidence the influence of statisticians in recent methodological advances throughout the bioinformatics pipeline and advocate for the expansion of this role. The latest acquisition platforms, such as next generation sequencing (genomics/transcriptomics) and hyphenated mass spectrometry (proteomics/metabolomics), output raw datasets in the order of gigabytes; it is not unusual to acquire a terabyte or more of data per study. The increasing computational burden this brings is a further impediment against the use of statistically rigorous methodology in the pre-processing stages of the bioinformatics pipeline. In this discussion I describe the mass spectrometry pipeline and use it as an example to show that beneath this challenge lies a two-fold opportunity: (a) Biological complexity and dynamic range is still well beyond what is captured by current processing methodology; hence, potential biomarkers and mechanistic insights are consistently missed; (b) Statistical science could play a larger role in optimizing the acquisition process itself. Data rates will continue to increase as routine clinical omics analysis moves to large-scale facilities with systematic, standardized protocols. Key inferential gains will be achieved by borrowing strength across the sum total of all analyzed studies, a task best underpinned by appropriate statistical modelling.

Statistical control of peptide and protein error rates in large-scale targeted data-independent acquisition analyses

Liquid chromatography coupled to tandem mass spectrometry is the main method for high-throughput identification and quantification of peptides and inferred proteins. Within this field, data-independent acquisition (DIA) combined with peptide-centric scoring, exemplified by SWATH-MS, emerged as a scalable method to achieve deep and consistent proteome coverage across large-scale datasets. Here we discuss the adaptation of statistical concepts developed for discovery proteomics based on spectrum-centric scoring to large-scale DIA experiments analyzed with peptide-centric scoring strategies and provide guidance on their application. We show that optimal tradeoffs between sensitivity and specificity require careful considerations of the relationship between proteins in the samples and proteins represented in the spectral library. We propose the application of a global analyte constraint to prevent accumulation of false positives across large-scale datasets. Furthermore, to increase the quality and reproducibility of published proteomic results, well-established confidence criteria should be reported for detected peptide queries, peptides and inferred proteins.

Proteomic differences in amyloid plaques in rapidly progressive and sporadic Alzheimer's disease

Rapidly progressive Alzheimer's disease (rpAD) is a particularly aggressive form of Alzheimer's disease, with a median survival time of 7-10 months after diagnosis. Why these patients have such a rapid progression of Alzheimer's disease is currently unknown. To further understand pathological differences between rpAD and typical sporadic Alzheimer's disease (sAD) we used localized proteomics to analyze the protein differences in amyloid plaques in rpAD and sAD. Label-free quantitative LC-MS/MS was performed on amyloid plaques microdissected from rpAD and sAD patients (n = 22 for each patient group) and protein expression differences were quantified. On average, 913 ± 30 (mean ± SEM) proteins were quantified in plaques from each patient and 279 of these proteins were consistently found in plaques from every patient. We found significant differences in protein composition between rpAD and sAD plaques. We found that rpAD plaques contained significantly higher levels of neuronal proteins (p = 0.0017) and significantly lower levels of astrocytic proteins (p = 1.08 × 10^-6). Unexpectedly, cumulative protein differences in rpAD plaques did not suggest accelerated typical sAD. Plaques from patients with rpAD were particularly abundant in synaptic proteins, especially those involved in synaptic vesicle release, highlighting the potential importance of synaptic dysfunction in the accelerated development of plaque pathology in rpAD. Combined, our data provide new direct evidence that amyloid plaques do not all have the same protein composition and that the proteomic differences in plaques could provide important insight into the factors that contribute to plaque development. The cumulative protein differences in rpAD plaques suggest rpAD may be a novel subtype of Alzheimer's disease.

Covariation of Peptide Abundances Accurately Reflects Protein Concentration Differences

Most implementations of mass spectrometry-based proteomics involve enzymatic digestion of proteins, expanding the analysis to multiple proteolytic peptides for each protein. Currently, there is no consensus of how to summarize peptides' abundances to protein concentrations, and such efforts are complicated by the fact that error control normally is applied to the identification process, and do not directly control errors linking peptide abundance measures to protein concentration. Peptides resulting from suboptimal digestion or being partially modified are not representative of the protein concentration. Without a mechanism to remove such unrepresentative peptides, their abundance adversely impacts the estimation of their protein's concentration. Here, we present a relative quantification approach, Diffacto, that applies factor analysis to extract the covariation of peptides' abundances. The method enables a weighted geometrical average summarization and automatic elimination of incoherent peptides. We demonstrate, based on a set of controlled label-free experiments using standard mixtures of proteins, that the covariation structure extracted by the factor analysis accurately reflects protein concentrations. In the 1% peptide-spectrum match-level FDR data set, as many as 11% of the peptides have abundance differences incoherent with the other peptides attributed to the same protein. If not controlled, such contradicting peptide abundance have a severe impact on protein quantifications. When adding the quantities of each protein's three most abundant peptides, we note as many as 14% of the proteins being estimated as having a negative correlation with their actual concentration differences between samples. Diffacto reduced the amount of such obviously incorrectly quantified proteins to 1.6%. Furthermore, by analyzing clinical data sets from two breast cancer studies, our method revealed the persistent proteomic signatures linked to three subtypes of breast cancer. We conclude that Diffacto can facilitate the interpretation and enhance the utility of most types of proteomics data.

Fast and Accurate Protein False Discovery Rates on Large-Scale Proteomics Data Sets with Percolator 3.0

Percolator is a widely used software tool that increases yield in shotgun proteomics experiments and assigns reliable statistical confidence measures, such as q values and posterior error probabilities, to peptides and peptide-spectrum matches (PSMs) from such experiments. Percolator's processing speed has been sufficient for typical data sets consisting of hundreds of thousands of PSMs. With our new scalable approach, we can now also analyze millions of PSMs in a matter of minutes on a commodity computer. Furthermore, with the increasing awareness for the need for reliable statistics on the protein level, we compared several easy-to-understand protein inference methods and implemented the best-performing method-grouping proteins by their corresponding sets of theoretical peptides and then considering only the best-scoring peptide for each protein-in the Percolator package. We used Percolator 3.0 to analyze the data from a recent study of the draft human proteome containing 25 million spectra (PM:24870542). The source code and Ubuntu, Windows, MacOS, and Fedora binary packages are available from http://percolator.ms/ under an Apache 2.0 license. Graphical Abstract ᅟ.

How to talk about protein-level false discovery rates in shotgun proteomics

A frequently sought output from a shotgun proteomics experiment is a list of proteins that we believe to have been present in the analyzed sample before proteolytic digestion. The standard technique to control for errors in such lists is to enforce a preset threshold for the false discovery rate (FDR). Many consider protein-level FDRs a difficult and vague concept, as the measurement entities, spectra, are manifestations of peptides and not proteins. Here, we argue that this confusion is unnecessary and provide a framework on how to think about protein-level FDRs, starting from its basic principle: the null hypothesis. Specifically, we point out that two competing null hypotheses are used concurrently in today's protein inference methods, which has gone unnoticed by many. Using simulations of a shotgun proteomics experiment, we show how confusing one null hypothesis for the other can lead to serious discrepancies in the FDR. Furthermore, we demonstrate how the same simulations can be used to verify FDR estimates of protein inference methods. In particular, we show that, for a simple protein inference method, decoy models can be used to accurately estimate protein-level FDRs for both competing null hypotheses.

Full-Featured Search Algorithm for Negative Electron-Transfer Dissociation

Negative electron-transfer dissociation (NETD) has emerged as a premier tool for peptide anion analysis, offering access to acidic post-translational modifications and regions of the proteome that are intractable with traditional positive-mode approaches. Whole-proteome scale characterization is now possible with NETD, but proper informatic tools are needed to capitalize on advances in instrumentation. Currently only one database search algorithm (OMSSA) can process NETD data. Here we implement NETD search capabilities into the Byonic platform to improve the sensitivity of negative-mode data analyses, and we benchmark these improvements using 90 min LC-MS/MS analyses of tryptic peptides from human embryonic stem cells. With this new algorithm for searching NETD data, we improved the number of successfully identified spectra by as much as 80% and identified 8665 unique peptides, 24 639 peptide spectral matches, and 1338 proteins in activated-ion NETD analyses, more than doubling identifications from previous negative-mode characterizations of the human proteome. Furthermore, we reanalyzed our recently published large-scale, multienzyme negative-mode yeast proteome data, improving peptide and peptide spectral match identifications and considerably increasing protein sequence coverage. In all, we show that new informatics tools, in combination with recent advances in data acquisition, can significantly improve proteome characterization in negative-mode approaches.

Comparative Analysis of Label-Free and 8-Plex iTRAQ Approach for Quantitative Tissue Proteomic Analysis

High resolution proteomics approaches have been successfully utilized for the comprehensive characterization of the cell proteome. However, in the case of quantitative proteomics an open question still remains, which quantification strategy is best suited for identification of biologically relevant changes, especially in clinical specimens. In this study, a thorough comparison of a label-free approach (intensity-based) and 8-plex iTRAQ was conducted as applied to the analysis of tumor tissue samples from non-muscle invasive and muscle-invasive bladder cancer. For the latter, two acquisition strategies were tested including analysis of unfractionated and fractioned iTRAQ-labeled peptides. To reduce variability, aliquots of the same protein extract were used as starting material, whereas to obtain representative results per method further sample processing and MS analysis were conducted according to routinely applied protocols. Considering only multiple-peptide identifications, LC-MS/MS analysis resulted in the identification of 910, 1092 and 332 proteins by label-free, fractionated and unfractionated iTRAQ, respectively. The label-free strategy provided higher protein sequence coverage compared to both iTRAQ experiments. Even though pre-fraction of the iTRAQ labeled peptides allowed for a higher number of identifications, this was not accompanied by a respective increase in the number of differentially expressed changes detected. Validity of the proteomics output related to protein identification and differential expression was determined by comparison to existing data in the field (Protein Atlas and published data on the disease). All methods predicted changes which to a large extent agreed with published data, with label-free providing a higher number of significant changes than iTRAQ. Conclusively, both label-free and iTRAQ (when combined to peptide fractionation) provide high proteome coverage and apparently valid predictions in terms of differential expression, nevertheless label-free provides higher sequence coverage and ultimately detects a higher number of differentially expressed proteins. The risk for receiving false associations still exists, particularly when analyzing highly heterogeneous biological samples, raising the need for the analysis of higher sample numbers and/or application of adjustment for multiple testing.

Comparison of Depletion Strategies for the Enrichment of Low-Abundance Proteins in Urine

Proteome analysis of complex biological samples for biomarker identification remains challenging, among others due to the extended range of protein concentrations. High-abundance proteins like albumin or IgG of plasma and urine, may interfere with the detection of potential disease biomarkers. Currently, several options are available for the depletion of abundant proteins in plasma. However, the applicability of these methods in urine has not been thoroughly investigated. In this study, we compared different, commercially available immunodepletion and ion-exchange based approaches on urine samples from both healthy subjects and CKD patients, for their reproducibility and efficiency in protein depletion. A starting urine volume of 500 μL was used to simulate conditions of a multi-institutional biomarker discovery study. All depletion approaches showed satisfactory reproducibility (n=5) in protein identification as well as protein abundance. Comparison of the depletion efficiency between the unfractionated and fractionated samples and the different depletion strategies, showed efficient depletion in all cases, with the exception of the ion-exchange kit. The depletion efficiency was found slightly higher in normal than in CKD samples and normal samples yielded more protein identifications than CKD samples when using both initial as well as corresponding depleted fractions. Along these lines, decrease in the amount of albumin and other targets as applicable, following depletion, was observed. Nevertheless, these depletion strategies did not yield a higher number of identifications in neither the urine from normal nor CKD patients. Collectively, when analyzing urine in the context of CKD biomarker identification, no added value of depletion strategies can be observed and analysis of unfractionated starting urine appears to be preferable.

A Fast Numerical Method for Max-Convolution and the Application to Efficient Max-Product Inference in Bayesian Networks

Observations depending on sums of random variables are common throughout many fields; however, no efficient solution is currently known for performing max-product inference on these sums of general discrete distributions (max-product inference can be used to obtain maximum a posteriori estimates). The limiting step to max-product inference is the max-convolution problem (sometimes presented in log-transformed form and denoted as "infimal convolution," "min-convolution," or "convolution on the tropical semiring"), for which no O(k log(k)) method is currently known. Presented here is an O(k log(k)) numerical method for estimating the max-convolution of two nonnegative vectors (e.g., two probability mass functions), where k is the length of the larger vector. This numerical max-convolution method is then demonstrated by performing fast max-product inference on a convolution tree, a data structure for performing fast inference given information on the sum of n discrete random variables in O(nk log(nk)log(n)) steps (where each random variable has an arbitrary prior distribution on k contiguous possible states). The numerical max-convolution method can be applied to specialized classes of hidden Markov models to reduce the runtime of computing the Viterbi path from nk(2) to nk log(k), and has potential application to the all-pairs shortest paths problem.

Bayesian proteoform modeling improves protein quantification of global proteomic measurements

As the capability of mass spectrometry-based proteomics has matured, tens of thousands of peptides can be measured simultaneously, which has the benefit of offering a systems view of protein expression. However, a major challenge is that, with an increase in throughput, protein quantification estimation from the native measured peptides has become a computational task. A limitation to existing computationally driven protein quantification methods is that most ignore protein variation, such as alternate splicing of the RNA transcript and post-translational modifications or other possible proteoforms, which will affect a significant fraction of the proteome. The consequence of this assumption is that statistical inference at the protein level, and consequently downstream analyses, such as network and pathway modeling, have only limited power for biomarker discovery. Here, we describe a Bayesian Proteoform Quantification model (BP-Quant)(1) that uses statistically derived peptides signatures to identify peptides that are outside the dominant pattern or the existence of multiple overexpressed patterns to improve relative protein abundance estimates. It is a research-driven approach that utilizes the objectives of the experiment, defined in the context of a standard statistical hypothesis, to identify a set of peptides exhibiting similar statistical behavior relating to a protein. This approach infers that changes in relative protein abundance can be used as a surrogate for changes in function, without necessarily taking into account the effect of differential post-translational modifications, processing, or splicing in altering protein function. We verify the approach using a dilution study from mouse plasma samples and demonstrate that BP-Quant achieves similar accuracy as the current state-of-the-art methods at proteoform identification with significantly better specificity. BP-Quant is available as a MatLab® and R packages.

Improving protein identification from tandem mass spectrometry data by one-step methods and integrating data from other platforms

MOTIVATION

Many approaches have been proposed for the protein identification problem based on tandem mass spectrometry (MS/MS) data. In these experiments, proteins are digested into peptides and the resulting peptide mixture is subjected to mass spectrometry. Some interesting putative peptide features (peaks) are selected from the mass spectra. Following that, the precursor ions undergo fragmentation and are analyzed by MS/MS. The process of identification of peptides from the mass spectra and the constituent proteins in the sample is called protein identification from MS/MS data. There are many two-step protein identification procedures, reviewed in the literature, which first attempt to identify the peptides in a separate process and then use these results to infer the proteins. However, in recent years, there have been attempts to provide a one-step solution to protein identification, which simultaneously identifies the proteins and the peptides in the sample.

RESULTS

In this review, we briefly introduce the most popular two-step protein identification procedure, PeptideProphet coupled with ProteinProphet. Following that, we describe the difficulties with two-step procedures and review some recently introduced one-step protein/peptide identification procedures that do not suffer from these issues. The focus of this review is on one-step procedures that are based on statistical likelihood-based models, but some discussion of other one-step procedures is also included. We report comparative performances of one-step and two-step methods, which support the overall superiorities of one-step procedures. We also cover some recent efforts to improve protein identification by incorporating other molecular data along with MS/MS data.

A Scalable Approach for Protein False Discovery Rate Estimation in Large Proteomic Data Sets

Calculating the number of confidently identified proteins and estimating false discovery rate (FDR) is a challenge when analyzing very large proteomic data sets such as entire human proteomes. Biological and technical heterogeneity in proteomic experiments further add to the challenge and there are strong differences in opinion regarding the conceptual validity of a protein FDR and no consensus regarding the methodology for protein FDR determination. There are also limitations inherent to the widely used classic target-decoy strategy that particularly show when analyzing very large data sets and that lead to a strong over-representation of decoy identifications. In this study, we investigated the merits of the classic, as well as a novel target-decoy-based protein FDR estimation approach, taking advantage of a heterogeneous data collection comprised of ∼19,000 LC-MS/MS runs deposited in ProteomicsDB (https://www.proteomicsdb.org). The "picked" protein FDR approach treats target and decoy sequences of the same protein as a pair rather than as individual entities and chooses either the target or the decoy sequence depending on which receives the highest score. We investigated the performance of this approach in combination with q-value based peptide scoring to normalize sample-, instrument-, and search engine-specific differences. The "picked" target-decoy strategy performed best when protein scoring was based on the best peptide q-value for each protein yielding a stable number of true positive protein identifications over a wide range of q-value thresholds. We show that this simple and unbiased strategy eliminates a conceptual issue in the commonly used "classic" protein FDR approach that causes overprediction of false-positive protein identification in large data sets. The approach scales from small to very large data sets without losing performance, consistently increases the number of true-positive protein identifications and is readily implemented in proteomics analysis software.

Mass spectrometry-based protein identification with accurate statistical significance assignment

MOTIVATION

Assigning statistical significance accurately has become increasingly important as metadata of many types, often assembled in hierarchies, are constructed and combined for further biological analyses. Statistical inaccuracy of metadata at any level may propagate to downstream analyses, undermining the validity of scientific conclusions thus drawn. From the perspective of mass spectrometry-based proteomics, even though accurate statistics for peptide identification can now be achieved, accurate protein level statistics remain challenging.

RESULTS

We have constructed a protein ID method that combines peptide evidences of a candidate protein based on a rigorous formula derived earlier; in this formula the database P-value of every peptide is weighted, prior to the final combination, according to the number of proteins it maps to. We have also shown that this protein ID method provides accurate protein level E-value, eliminating the need of using empirical post-processing methods for type-I error control. Using a known protein mixture, we find that this protein ID method, when combined with the Sorić formula, yields accurate values for the proportion of false discoveries. In terms of retrieval efficacy, the results from our method are comparable with other methods tested.

AVAILABILITY AND IMPLEMENTATION

The source code, implemented in C++ on a linux system, is available for download at ftp://ftp.ncbi.nlm.nih.gov/pub/qmbp/qmbp_ms/RAId/RAId_Linux_64Bit.

The probabilistic convolution tree: efficient exact Bayesian inference for faster LC-MS/MS protein inference

Exact Bayesian inference can sometimes be performed efficiently for special cases where a function has commutative and associative symmetry of its inputs (called “causal independence”). For this reason, it is desirable to exploit such symmetry on big data sets. Here we present a method to exploit a general form of this symmetry on probabilistic adder nodes by transforming those probabilistic adder nodes into a probabilistic convolution tree with which dynamic programming computes exact probabilities. A substantial speedup is demonstrated using an illustration example that can arise when identifying splice forms with bottom-up mass spectrometry-based proteomics. On this example, even state-of-the-art exact inference algorithms require a runtime more than exponential in the number of splice forms considered. By using the probabilistic convolution tree, we reduce the runtime to and the space to where is the number of variables joined by an additive or cardinal operator. This approach, which can also be used with junction tree inference, is applicable to graphs with arbitrary dependency on counting variables or cardinalities and can be used on diverse problems and fields like forward error correcting codes, elemental decomposition, and spectral demixing. The approach also trivially generalizes to multiple dimensions.

Machine learning applications in proteomics research: how the past can boost the future

Machine learning is a subdiscipline within artificial intelligence that focuses on algorithms that allow computers to learn solving a (complex) problem from existing data. This ability can be used to generate a solution to a particularly intractable problem, given that enough data are available to train and subsequently evaluate an algorithm on. Since MS-based proteomics has no shortage of complex problems, and since publicly available data are becoming available in ever growing amounts, machine learning is fast becoming a very popular tool in the field. We here therefore present an overview of the different applications of machine learning in proteomics that together cover nearly the entire wet- and dry-lab workflow, and that address key bottlenecks in experiment planning and design, as well as in data processing and analysis.

A combinatorial perspective of the protein inference problem

In a shotgun proteomics experiment, proteins are the most biologically meaningful output. The success of proteomics studies depends on the ability to accurately and efficiently identify proteins. Many methods have been proposed to facilitate the identification of proteins from peptide identification results. However, the relationship between protein identification and peptide identification has not been thoroughly explained before. In this paper, we devote ourselves to a combinatorial perspective of the protein inference problem. We employ combinatorial mathematics to calculate the conditional protein probabilities (protein probability means the probability that a protein is correctly identified) under three assumptions, which lead to a lower bound, an upper bound, and an empirical estimation of protein probabilities, respectively. The combinatorial perspective enables us to obtain an analytical expression for protein inference. Our method achieves comparable results with ProteinProphet in a more efficient manner in experiments on two data sets of standard protein mixtures and two data sets of real samples. Based on our model, we study the impact of unique peptides and degenerate peptides (degenerate peptides are peptides shared by at least two proteins) on protein probabilities. Meanwhile, we also study the relationship between our model and ProteinProphet. We name our program ProteinInfer. Its Java source code, our supplementary document and experimental results are available at: >http://bioinformatics.ust.hk/proteininfer.

Nonparametric Bayesian evaluation of differential protein quantification

Arbitrary cutoffs are ubiquitous in quantitative computational proteomics: maximum acceptable MS/MS PSM or peptide q value, minimum ion intensity to calculate a fold change, the minimum number of peptides that must be available to trust the estimated protein fold change (or the minimum number of PSMs that must be available to trust the estimated peptide fold change), and the "significant" fold change cutoff. Here we introduce a novel experimental setup and nonparametric Bayesian algorithm for determining the statistical quality of a proposed differential set of proteins or peptides. By comparing putatively nonchanging case-control evidence to an empirical null distribution derived from a control-control experiment, we successfully avoid some of these common parameters. We then apply our method to evaluating different fold-change rules and find that for our data a 1.2-fold change is the most permissive of the plausible fold-change rules.

Non-canonical ubiquitylation: mechanisms and consequences

Post-translational protein modifications initiate, regulate, propagate and terminate a wide variety of processes in cells, and in particular, ubiquitylation targets substrate proteins for degradation, subcellular translocation, cell signaling and multiple other cellular events. Modification of substrate proteins is widely observed to occur via covalent linkages of ubiquitin to the amine groups of lysine side-chains. However, in recent years several new modes of ubiquitin chain attachment have emerged. For instance, covalent modification of non-lysine sites in substrate proteins is theoretically possible according to basic chemical principles underlying the ubiquitylation process, and evidence is building that sites such as the N-terminal amine group of a protein, the hydroxyl group of serine and threonine residues and even the thiol groups of cysteine residues are all employed as sites of ubiquitylation. However, the potential importance of this "non-canonical ubiquitylation" of substrate proteins on sites other than lysine residues has been largely overlooked. This review aims to highlight the unusual features of the process of non-canonical ubiquitylation and the consequences of these events on the activity and fate of a protein.

A non-parametric cutout index for robust evaluation of identified proteins

This paper proposes a novel, automated method for evaluating sets of proteins identified using mass spectrometry. The remaining peptide-spectrum match score distributions of protein sets are compared to an empirical absent peptide-spectrum match score distribution, and a Bayesian non-parametric method reminiscent of the Dirichlet process is presented to accurately perform this comparison. Thus, for a given protein set, the process computes the likelihood that the proteins identified are correctly identified. First, the method is used to evaluate protein sets chosen using different protein-level false discovery rate (FDR) thresholds, assigning each protein set a likelihood. The protein set assigned the highest likelihood is used to choose a non-arbitrary protein-level FDR threshold. Because the method can be used to evaluate any protein identification strategy (and is not limited to mere comparisons of different FDR thresholds), we subsequently use the method to compare and evaluate multiple simple methods for merging peptide evidence over replicate experiments. The general statistical approach can be applied to other types of data (e.g. RNA sequencing) and generalizes to multivariate problems.

Recognizing uncertainty increases robustness and reproducibility of mass spectrometry-based protein inferences

Parsimony and protein grouping are widely employed to enforce economy in the number of identified proteins, with the goal of increasing the quality and reliability of protein identifications; however, in a counterintuitive manner, parsimony and protein grouping may actually decrease the reproducibility and interpretability of protein identifications. We present a simple illustration demonstrating ways in which parsimony and protein grouping may lower the reproducibility or interpretability of results. We then provide an example of a data set where a probabilistic method increases the reproducibility and interpretability of identifications made on replicate analyses of Human Du145 prostate cancer cell lines.

Computational approaches to protein inference in shotgun proteomics

Shotgun proteomics has recently emerged as a powerful approach to characterizing proteomes in biological samples. Its overall objective is to identify the form and quantity of each protein in a high-throughput manner by coupling liquid chromatography with tandem mass spectrometry. As a consequence of its high throughput nature, shotgun proteomics faces challenges with respect to the analysis and interpretation of experimental data. Among such challenges, the identification of proteins present in a sample has been recognized as an important computational task. This task generally consists of (1) assigning experimental tandem mass spectra to peptides derived from a protein database, and (2) mapping assigned peptides to proteins and quantifying the confidence of identified proteins. Protein identification is fundamentally a statistical inference problem with a number of methods proposed to address its challenges. In this review we categorize current approaches into rule-based, combinatorial optimization and probabilistic inference techniques, and present them using integer programming and Bayesian inference frameworks. We also discuss the main challenges of protein identification and propose potential solutions with the goal of spurring innovative research in this area.