On the excessive use of coefficient of variation as a metric of quantitation quality in proteomics.

The coefficient of variation (CV) is often used in proteomics as a proxy to characterize the performance of a quantitation method and/or the related software. In this note, we question the excessive reliance on this metric in quantitative proteomics that may result in erroneous conclusions. We support this note using a ground-truth Human-Yeast-E. coli dataset demonstrating in a number of cases that erroneous data processing methods may lead to a low CV which has nothing to do with these methods' performances in quantitation.

Proteomics-based scoring of cellular response to stimuli for improved characterization of signaling pathway activity.

Omics technologies focus on uncovering the complex nature of molecular mechanisms in cells and organisms, including biomarkers and drug targets discovery. Aiming at these tasks, we see that information extracted from omics data is still underused. In particular, characteristics of differentially regulated molecules can be combined in a single score to quantify the signaling pathway activity. Such a metric can be useful for comprehensive data interpretation to follow: (1) developing molecular responses in time; (2) potency of a drug on a certain cell culture; (3) ranking the signaling pathway activity in stimulated cells; and (4) integration of the omics data and assay-based measurements of cell viability, cytotoxicity, and proliferation. With recent advances in ultrafast mass spectrometry for quantitative proteomics allowing data collection in a few minutes, proteomics score for cellular response to stimuli can become a fast, accurate, and informative complement to bioassays. Here, we utilized an interquartile-based selection of differentially regulated features and a variety of schemes for quantifying cellular responses to come up with the quantitative metric for total cellular response and pathway activity. Validation was performed using antiproliferative and virus assays and label-free proteomics data collected for cancer cells subjected to drug stimulation.

IQMMA: Efficient MS1 Intensity Extraction Pipeline Using Multiple Feature Detection Algorithms for DDA Proteomics.

One of the key steps in data dependent acquisition (DDA) proteomics is detection of peptide isotopic clusters, also called "features", in MS1 spectra and matching them to MS/MS-based peptide identifications. A number of peptide feature detection tools became available in recent years, each relying on its own matching algorithm. Here, we provide an integrated solution, the intensity-based Quantitative Mix and Match Approach (IQMMA), which integrates a number of untargeted peptide feature detection algorithms and returns the most probable intensity values for the MS/MS-based identifications. IQMMA was tested using available proteomic data acquired for both well-characterized (ground truth) and real-world biological samples, including a mix of Yeast and E. coli digests spiked at different concentrations into the Human K562 digest used as a background, and a set of glioblastoma cell lines. Three open-source feature detection algorithms were integrated: Dinosaur, biosaur2, and OpenMS FeatureFinder. None of them was found optimal when applied individually to all the data sets employed in this work; however, their combined use in IQMMA improved efficiency of subsequent protein quantitation. The software implementing IQMMA is freely available at https://github.com/PostoenkoVI/IQMMA under Apache 2.0 license.

Massive Proteogenomic Reanalysis of Publicly Available Proteomic Datasets of Human Tissues in Search for Protein Recoding via Adenosine-to-Inosine RNA Editing.

The proteogenomic search pipeline developed in this work has been applied for reanalysis of 40 publicly available shotgun proteomic datasets from various human tissues comprising more than 8000 individual LC-MS/MS runs, of which 5442 .raw data files were processed in total. This reanalysis was focused on searching for ADAR-mediated RNA editing events, their clustering across samples of different origins, and classification. In total, 33 recoded protein sites were identified in 21 datasets. Of those, 18 sites were detected in at least two datasets, representing the core human protein editome. In agreement with prior artworks, neural and cancer tissues were found to be enriched with recoded proteins. Quantitative analysis indicated that recoding the rate of specific sites did not directly depend on the levels of ADAR enzymes or targeted proteins themselves, rather it was governed by differential and yet undescribed regulation of interaction of enzymes with mRNA. Nine recoding sites conservative between humans and rodents were validated by targeted proteomics using stable isotope standards in the murine brain cortex and cerebellum, and an additional one was validated in human cerebrospinal fluid. In addition to previous data of the same type from cancer proteomes, we provide a comprehensive catalog of recoding events caused by ADAR RNA editing in the human proteome.

Identification of Alternative Splicing in Proteomes of Human Melanoma Cell Lines without RNA Sequencing Data.

Alternative splicing is one of the main regulation pathways in living cells beyond simple changes in the level of protein expression. Most of the approaches proposed in proteomics for the identification of specific splicing isoforms require a preliminary deep transcriptomic analysis of the sample under study, which is not always available, especially in the case of the re-analysis of previously acquired data. Herein, we developed new algorithms for the identification and validation of protein splice isoforms in proteomic data in the absence of RNA sequencing of the samples under study. The bioinformatic approaches were tested on the results of proteome analysis of human melanoma cell lines, obtained earlier by high-resolution liquid chromatography and mass spectrometry (LC-MS). A search for alternative splicing events for each of the cell lines studied was performed against the database generated from all known transcripts (RefSeq) and the one composed of peptide sequences, which included all biologically possible combinations of exons. The identifications were filtered using the prediction of both retention times and relative intensities of fragment ions in the corresponding mass spectra. The fragmentation mass spectra corresponding to the discovered alternative splicing events were additionally examined for artifacts. Selected splicing events were further validated at the mRNA level by quantitative PCR.

Validating Amino Acid Variants in Proteogenomics Using Sequence Coverage by Multiple Reads.

Mass spectrometry-based proteome analysis implies matching the mass spectra of proteolytic peptides to amino acid sequences predicted from genomic sequences. Reliability of peptide variant identification in proteogenomic studies is often lacking. We propose a way to interpret shotgun proteomics results, specifically in the data-dependent acquisition mode, as protein sequence coverage by multiple reads as it is done in nucleic acid sequencing for calling of single nucleotide variants. Multiple reads for each sequence position could be provided by overlapping distinct peptides, thus confirming the presence of certain amino acid residues in the overlapping stretch with a lower false discovery rate. Overlapping distinct peptides originate from miscleaved tryptic peptides in combination with their properly cleaved counterparts and from peptides generated by multiple proteases after the same specimen is subject to parallel digestion and analyzed separately. We illustrate this approach using publicly available multiprotease data sets and our own data generated for the HEK-293 cell line digests obtained using trypsin, LysC, and GluC proteases. Totally, up to 30% of the whole proteome was covered by tryptic peptides with up to 7% covered twofold and more. The proteogenomic analysis of the HEK-293 cell line revealed 36 single amino acid variants, seven of which were supported by multiple reads.

A Comprehensive Evaluation of Consensus Spectrum Generation Methods in Proteomics.

Spectrum clustering is a powerful strategy to minimize redundant mass spectra by grouping them based on similarity, with the aim of forming groups of mass spectra from the same repeatedly measured analytes. Each such group of near-identical spectra can be represented by its so-called consensus spectrum for downstream processing. Although several algorithms for spectrum clustering have been adequately benchmarked and tested, the influence of the consensus spectrum generation step is rarely evaluated. Here, we present an implementation and benchmark of common consensus spectrum algorithms, including spectrum averaging, spectrum binning, the most similar spectrum, and the best-identified spectrum. We have analyzed diverse public data sets using two different clustering algorithms (spectra-cluster and MaRaCluster) to evaluate how the consensus spectrum generation procedure influences downstream peptide identification. The BEST and BIN methods were found the most reliable methods for consensus spectrum generation, including for data sets with post-translational modifications (PTM) such as phosphorylation. All source code and data of the present study are freely available on GitHub at https://github.com/statisticalbiotechnology/representative-spectra-benchmark.

DirectMS1Quant: Ultrafast Quantitative Proteomics with MS/MS-Free Mass Spectrometry.

Recently, we presented the DirectMS1 method of ultrafast proteome-wide analysis based on minute-long LC gradients and MS1-only mass spectra acquisition. Currently, the method provides the depth of human cell proteome coverage of 2500 proteins at a 1% false discovery rate (FDR) when using 5 min LC gradients and 7.3 min runtime in total. While the standard MS/MS approaches provide 4000-5000 protein identifications within a couple of hours of instrumentation time, we advocate here that the higher number of identified proteins does not always translate into better quantitation quality of the proteome analysis. To further elaborate on this issue, we performed a one-on-one comparison of quantitation results obtained using DirectMS1 with three popular MS/MS-based quantitation methods: label-free (LFQ) and tandem mass tag quantitation (TMT), both based on data-dependent acquisition (DDA) and data-independent acquisition (DIA). For comparison, we performed a series of proteome-wide analyses of well-characterized (ground truth) and biologically relevant samples, including a mix of UPS1 proteins spiked at different concentrations into an Echerichia coli digest used as a background and a set of glioblastoma cell lines. MS1-only data was analyzed using a novel quantitation workflow called DirectMS1Quant developed in this work. The results obtained in this study demonstrated comparable quantitation efficiency of 5 min DirectMS1 with both TMT and DIA methods, yet the latter two utilized a 10-20-fold longer instrumentation time.

Proteomic Analysis of Zebrafish Protein Recoding via mRNA Editing by ADAR Enzymes.

RNA editing by adenosine deaminases of the ADAR family can lead to protein recoding, since inosine formed from adenosine in mRNA is complementary to cytosine; the resulting codon editing might introduce amino acid substitutions into translated proteins. Proteome recoding can have functional consequences which have been described in many animals including humans. Using protein recoding database derived from publicly available transcriptome data, we identified for the first time the recoding sites in the zebrafish shotgun proteomes. Out of more than a hundred predicted recoding events, ten substitutions were found in six used datasets. Seven of them were in the AMPA glutamate receptor subunits, whose recoding has been well described, and are conserved among vertebrates. Three sites were specific for zebrafish proteins and were found in the transmembrane receptors astrotactin 1 and neuregulin 3b (proteins involved in the neuronal adhesion and signaling) and in the rims2b gene product (presynaptic membrane protein participating in the neurotransmitter release), respectively. Further studies are needed to elucidate the role of recoding of the said three proteins in the zebrafish.

Proteome-Wide Analysis of ADAR-Mediated Messenger RNA Editing during Fruit Fly Ontogeny.

Adenosine-to-inosine RNA editing is an enzymatic post-transcriptional modification which modulates immunity and neural transmission in multicellular organisms. In particular, it involves editing of mRNA codons with the resulting amino acid substitutions. We identified such sites for developmental proteomes of Drosophila melanogaster at the protein level using available data for 15 stages of fruit fly development from egg to imago and 14 time points of embryogenesis. In total, 40 sites were obtained, each belonging to a unique protein, including four sites related to embryogenesis. The interactome analysis has revealed that the majority of the editing-recoded proteins were associated with synaptic vesicle trafficking and actomyosin organization. Quantitation data analysis suggested the existence of a phase-specific RNA editing regulation with yet unknown mechanisms. These findings supported the transcriptome analysis results, which showed that a burst in the RNA editing occurs during insect metamorphosis from pupa to imago. Finally, targeted proteomic analysis was performed to quantify editing-recoded and genomically encoded versions of five proteins in brains of larvae, pupae, and imago insects, which showed a clear tendency toward an increase in the editing rate for each of them. These results will allow a better understanding of the protein role in physiological effects of RNA editing.

DirectMS1: MS/MS-Free Identification of 1000 Proteins of Cellular Proteomes in 5 Minutes.

Proteome characterization relies heavily on tandem mass spectrometry (MS/MS) and is thus associated with instrumentation complexity, lengthy analysis time, and limited duty cycle. It was always tempting to implement approaches that do not require MS/MS, yet they were constantly failing to achieve a meaningful depth of quantitative proteome coverage within short experimental times, which is particularly important for clinical or biomarker-discovery applications. Here, we report on the first successful attempt to develop a truly MS/MS-free method, DirectMS1, for bottom-up proteomics. The method is compared with the standard MS/MS-based data-dependent acquisition approach for proteome-wide analysis using 5 min LC gradients. Specifically, we demonstrate identification of 1 000 protein groups for a standard HeLa cell line digest. The amount of loaded sample was varied in a range from 1 to 500 ng, and the method demonstrated 10-fold higher sensitivity. Combined with the recently introduced Diffacto approach for relative protein quantification, DirectMS1 outperforms most popular MS/MS-based label-free quantitation approaches because of significantly higher protein sequence coverage.

Scavager: A Versatile Postsearch Validation Algorithm for Shotgun Proteomics Based on Gradient Boosting.

Shotgun proteomics workflows for database protein identification typically include a combination of search engines and postsearch validation software based mostly on machine learning algorithms. Here, a new postsearch validation tool called Scavager employing CatBoost, an open-source gradient boosting library, which shows improved efficiency compared with the other popular algorithms, such as Percolator, PeptideProphet, and Q-ranker, is presented. The comparison is done using multiple data sets and search engines, including MSGF+, MSFragger, X!Tandem, Comet, and recently introduced IdentiPy. Implemented in Python programming language, Scavager is open-source and freely available at https://bitbucket.org/markmipt/scavager.

Adenosine-to-Inosine RNA Editing in Mouse and Human Brain Proteomes.

Proteogenomics is based on the use of customized genome or RNA sequencing databases for interrogation of shotgun proteomics data in search for proteome-level evidence of genome variations or RNA editing. In this work, the products of adenosine-to-inosine RNA editing in human and murine brain proteomes are identified using publicly available brain proteome LC-MS/MS datasets and an RNA editome database compiled from several sources. After filtering of false-positive results, 20 and 37 sites of editing in proteins belonging to 14 and 32 genes are identified for murine and human brain proteomes, respectively. Eight sites of editing identified with high spectral counts overlapped between human and mouse brain samples. Some of these sites have been previously reported using orthogonal methods, such as α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) glutamate receptors, CYFIP2, coatomer alpha. Also, differential editing between neurons and microglia is demonstrated in this work for some of the proteins from primary murine brain cell cultures. Because many edited sites are still not characterized functionally at the protein level, the results provide a necessary background for their further analysis in normal and diseased cells and tissues using targeted proteomic approaches.

Pyteomics 4.0: Five Years of Development of a Python Proteomics Framework.

Many of the novel ideas that drive today's proteomic technologies are focused essentially on experimental or data-processing workflows. The latter are implemented and published in a number of ways, from custom scripts and programs, to projects built using general-purpose or specialized workflow engines; a large part of routine data processing is performed manually or with custom scripts that remain unpublished. Facilitating the development of reproducible data-processing workflows becomes essential for increasing the efficiency of proteomic research. To assist in overcoming the bioinformatics challenges in the daily practice of proteomic laboratories, 5 years ago we developed and announced Pyteomics, a freely available open-source library providing Python interfaces to proteomic data. We summarize the new functionality of Pyteomics developed during the time since its introduction.

Validation of Peptide Identification Results in Proteomics Using Amino Acid Counting.

The efficiency of proteome analysis depends strongly on the configuration parameters of the search engine. One of the murkiest and nontrivial among them is the list of amino acid modifications included for the search. Here, an approach called AA_stat is presented for uncovering the unexpected modifications of amino acid residues in the protein sequences, as well as possible artifacts of data acquisition or processing, in the results of proteome analyses. The approach is based on comparing the amino acid frequencies of different mass shifts observed using the open search method introduced recently. In this work, the proposed approach is applied to publicly available proteomic data is applied and its feasibility for discovering unaccounted modifications or possible pitfalls of the identification workflow is demonstrated. The algorithm is implemented in Python as an open-source command-line tool available at https://bitbucket.org/J_Bale/aa_stat/.

IdentiPy: An Extensible Search Engine for Protein Identification in Shotgun Proteomics.

We present an open-source, extensible search engine for shotgun proteomics. Implemented in Python programming language, IdentiPy shows competitive processing speed and sensitivity compared with the state-of-the-art search engines. It is equipped with a user-friendly web interface, IdentiPy Server, enabling the use of a single server installation accessed from multiple workstations. Using a simplified version of X!Tandem scoring algorithm and its novel "autotune" feature, IdentiPy outperforms the popular alternatives on high-resolution data sets. Autotune adjusts the search parameters for the particular data set, resulting in improved search efficiency and simplifying the user experience. IdentiPy with the autotune feature shows higher sensitivity compared with the evaluated search engines. IdentiPy Server has built-in postprocessing and protein inference procedures and provides graphic visualization of the statistical properties of the data set and the search results. It is open-source and can be freely extended to use third-party scoring functions or processing algorithms and allows customization of the search workflow for specialized applications.

FractionOptimizer: a method for optimal peptide fractionation in bottom-up proteomics.

Recent advances in mass spectrometry and separation technologies created the opportunities for deep proteome characterization using shotgun proteomics approaches. The "real world" sample complexity and high concentration range limit the sensitivity of this characterization. The common strategy for increasing the sensitivity is sample fractionation prior to analysis either at the protein or the peptide level. Typically, fractionation at the peptide level is performed using linear gradient high-performance liquid chromatography followed by uniform fraction collection. However, this way of peptide fractionation results in significantly suboptimal operation of the mass spectrometer due to the non-uniform distribution of peptides between the fractions. In this work, we propose an approach based on peptide retention time prediction allowing optimization of chromatographic conditions and fraction collection procedures. An open-source software implementing the approach called FractionOptimizer was developed and is available at http://hg.theorchromo.ru/FractionOptimizer . The performance of the developed tool was demonstrated for human embryonic kidney (HEK293) cell line lysate. In these experiments, we improved the uniformity of the peptides distribution between fractions. Moreover, in addition to 13,492 peptides, we found 6787 new peptides not identified in the experiments without fractionation and up to 800 new proteins (or 25%). Graphical abstract The analysis workflow employing FractionOptimizer software.

Unbiased False Discovery Rate Estimation for Shotgun Proteomics Based on the Target-Decoy Approach.

Target-decoy approach (TDA) is the dominant strategy for false discovery rate (FDR) estimation in mass-spectrometry-based proteomics. One of its main applications is direct FDR estimation based on counting of decoy matches above a certain score threshold. The corresponding equations are widely employed for filtering of peptide or protein identifications. In this work we consider a probability model describing the filtering process and find that, when decoy counting is used for q value estimation and subsequent filtering, a correction has to be introduced into these common equations for TDA-based FDR estimation. We also discuss the scale of variance of false discovery proportion (FDP) and propose using confidence intervals for more conservative FDP estimation in shotgun proteomics. The necessity of both the correction and the use of confidence intervals is especially pronounced when filtering small sets (such as in proteogenomics experiments) and when using very low FDR thresholds.

MS/MS-Free Protein Identification in Complex Mixtures Using Multiple Enzymes with Complementary Specificity.

In this work, we present the results of evaluation of a workflow that employs a multienzyme digestion strategy for MS1-based protein identification in "shotgun" proteomic applications. In the proposed strategy, several cleavage reagents of different specificity were used for parallel digestion of the protein sample followed by MS1 and retention time (RT) based search. Proof of principle for the proposed strategy was performed using experimental data obtained for the annotated 48-protein standard. By using the developed approach, up to 90% of proteins from the standard were unambiguously identified. The approach was further applied to HeLa proteome data. For the sample of this complexity, the proposed MS1-only strategy determined correctly up to 34% of all proteins identified using standard MS/MS-based database search. It was also found that the results of MS1-only search were independent of the chromatographic gradient time in a wide range of gradients from 15-120 min. Potentially, rapid MS1-only proteome characterization can be an alternative or complementary to the MS/MS-based "shotgun" analyses in the studies, in which the experimental time is more important than the depth of the proteome coverage.

Comparative evaluation of label-free quantification methods for shotgun proteomics.

RATIONALE: Label-free quantification (LFQ) is a popular strategy for shotgun proteomics. A variety of LFQ algorithms have been developed recently. However, a comprehensive comparison of the most commonly used LFQ methods is still rare, in part due to a lack of clear metrics for their evaluation and an annotated and quantitatively well-characterized data set. METHODS: Five LFQ methods were compared: spectral counting based algorithms SIN , emPAI, and NSAF, and approaches relying on the extracted ion chromatogram (XIC) intensities, MaxLFQ and Quanti. We used three criteria for performance evaluation: coefficient of variation (CV) of protein abundances between replicates; analysis of variance (ANOVA); and the root-mean-square error of logarithmized calculated concentration ratios, referred to as standard quantification error (SQE). Comparison was performed using a quantitatively annotated publicly available data set. RESULTS: The best results in terms of inter-replicate reproducibility were observed for MaxLFQ and NSAF, although they exhibited larger standard quantification errors. Using NSAF, all quantitatively annotated proteins were correctly identified in the Bonferronni-corrected results of the ANOVA test. SIN was found to be the most accurate in terms of SQE. Finally, the current implementations of XIC-based LFQ methods did not outperform the methods based on spectral counting for the data set used in this study. CONCLUSIONS: Surprisingly, the performances of XIC-based approaches measured using three independent metrics were found to be comparable with more straightforward and simple MS/MS-based spectral counting approaches. The study revealed no clear leader among the latter. Copyright © 2017 John Wiley & Sons, Ltd.