Koina: Democratizing machine learning for proteomics research.

Recent developments in machine-learning (ML) and deep-learning (DL) have immense potential for applications in proteomics, such as generating spectral libraries, improving peptide identification, and optimizing targeted acquisition modes. Although new ML/DL models for various applications and peptide properties are frequently published, the rate at which these models are adopted by the community is slow, which is mostly due to technical challenges. We believe that, for the community to make better use of state-of-the-art models, more attention should be spent on making models easy to use and accessible by the community. To facilitate this, we developed Koina, an open-source containerized, decentralized and online-accessible high-performance prediction service that enables ML/DL model usage in any pipeline. Using the widely used FragPipe computational platform as example, we show how Koina can be easily integrated with existing proteomics software tools and how these integrations improve data analysis.

Developmental stage-specific proteome analysis of the legume pod borer Maruca vitrata provides insights on relevant proteins.

The spotted pod borer, Maruca vitrata (Lepidoptera: Crambidae) is a destructive insect pest that inflicts significant productivity losses on important leguminous crops. Unravelling insect proteomes is vital to comprehend their fundamental molecular mechanisms. This research delved into the proteome profiles of four distinct stages -three larval and pupa of M. vitrata, utilizing LC-MS/MS label-free quantification-based methods. Employing comprehensive proteome analysis with fractionated datasets, we mapped 75 % of 3459 Drosophila protein orthologues out of which 2695 were identified across all developmental stages while, 137 and 94 were exclusive to larval and pupal stages respectively. Cluster analysis of 2248 protein orthologues derived from MaxQuant quantitative dataset depicted six clusters based on expression pattern similarity across stages. Consequently, gene ontology and protein-protein interaction network analyses using STRING database identified cluster 1 (58 proteins) and cluster 6 (25 proteins) associated with insect immune system and lipid metabolism. Furthermore, qRT-PCR-based expression analyses of ten selected proteins-coding genes authenticated the proteome data. Subsequently, functional validation of these chosen genes through gene silencing reduced their transcript abundance accompanied by a marked increase in mortality among dsRNA-injected larvae. Overall, this is a pioneering study to effectively develop a proteome atlas of M. vitrata as a potential resource for crop protection programs.

prolfqua: A Comprehensive R-Package for Proteomics Differential Expression Analysis.

Mass spectrometry is widely used for quantitative proteomics studies, relative protein quantification, and differential expression analysis of proteins. There is a large variety of quantification software and analysis tools. Nevertheless, there is a need for a modular, easy-to-use application programming interface in R that transparently supports a variety of well principled statistical procedures to make applying them to proteomics data, comparing and understanding their differences easy. The prolfqua package integrates essential steps of the mass spectrometry-based differential expression analysis workflow: quality control, data normalization, protein aggregation, statistical modeling, hypothesis testing, and sample size estimation. The package makes integrating new data formats easy. It can be used to model simple experimental designs with a single explanatory variable and complex experiments with multiple factors and hypothesis testing. The implemented methods allow sensitive and specific differential expression analysis. Furthermore, the package implements benchmark functionality that can help to compare data acquisition, data preprocessing, or data modeling methods using a gold standard data set. The application programmer interface of prolfqua strives to be clear, predictable, discoverable, and consistent to make proteomics data analysis application development easy and exciting. Finally, the prolfqua R-package is available on GitHub https://github.com/fgcz/prolfqua, distributed under the MIT license. It runs on all platforms supported by the R free software environment for statistical computing and graphics.

Bridging data management platforms and visualization tools to enable ad-hoc and smart analytics in life sciences.

Core facilities have to offer technologies that best serve the needs of their users and provide them a competitive advantage in research. They have to set up and maintain instruments in the range of ten to a hundred, which produce large amounts of data and serve thousands of active projects and customers. Particular emphasis has to be given to the reproducibility of the results. More and more, the entire process from building the research hypothesis, conducting the experiments, doing the measurements, through the data explorations and analysis is solely driven by very few experts in various scientific fields. Still, the ability to perform the entire data exploration in real-time on a personal computer is often hampered by the heterogeneity of software, the data structure formats of the output, and the enormous data sizes. These impact the design and architecture of the implemented software stack. At the Functional Genomics Center Zurich (FGCZ), a joint state-of-the-art research and training facility of ETH Zurich and the University of Zurich, we have developed the B-Fabric system, which has served for more than a decade, an entire life sciences community with fundamental data science support. In this paper, we sketch how such a system can be used to glue together data (including metadata), computing infrastructures (clusters and clouds), and visualization software to support instant data exploration and visual analysis. We illustrate our in-daily life implemented approach using visualization applications of mass spectrometry data.

Universal Spectrum Explorer: A Standalone (Web-)Application for Cross-Resource Spectrum Comparison.

Here, we present the Universal Spectrum Explorer (USE), a web-based tool based on IPSA for cross-resource (peptide) spectrum visualization and comparison (https://www.proteomicsdb.org/use/). Mass spectra under investigation can be either provided manually by the user (table format) or automatically retrieved from online repositories supporting access to spectral data via the universal spectrum identifier (USI), or requested from other resources and services implementing a newly designed REST interface. As a proof of principle, we implemented such an interface in ProteomicsDB thereby allowing the retrieval of spectra acquired within the ProteomeTools project or real-time prediction of tandem mass spectra from the deep learning framework Prosit. Annotated mirror spectrum plots can be exported from the USE as editable scalable high-quality vector graphics. The USE was designed and implemented with minimal external dependencies allowing local usage and integration into other web sites (https://github.com/kusterlab/universal_spectrum_explorer).

Quality standards in proteomics research facilities: Common standards and quality procedures are essential for proteomics facilities and their users.

Proteomics research infrastructures and core facilities within the Core for Life alliance advocate for community policies for quality control to ensure high standards in proteomics services.

The rawrr R Package: Direct Access to Orbitrap Data and Beyond.

The Bioconductor project (Nat. Methods2015, 12 (2), 115-121) has shown that the R statistical environment is a highly valuable tool for genomics data analysis, but with respect to proteomics, we are still missing low-level infrastructure to enable performant and robust analysis workflows in R. Fundamentally important are libraries that provide raw data access. Our R package rawDiag (J. Proteome Res.2018, 17 (8), 2908-2914) has provided the proof-of-principle how access to mass spectrometry raw files can be realized by wrapping a vendor-provided advanced programming interface (API) for the purpose of metadata analysis and visualization. Our novel package rawrr now provides complete, OS-independent access to all spectral data logged in Thermo Fisher Scientific raw files. In this technical note, we present implementation details and describe the main functionalities provided by the rawrr package. In addition, we report two use cases inspired by real-world research tasks that demonstrate the application of the package. The raw data used for demonstration purposes was deposited as MassIVE data set MSV000086542. Availability: https://github.com/fgcz/rawrr.

Gas-Phase Fragmentation of ADP-Ribosylated Peptides: Arginine-Specific Side-Chain Losses and Their Implication in Database Searches.

ADP-ribosylation is a reversible post-translational modification of proteins that has been linked to many biological processes. The identification of ADP-ribosylated proteins and particularly of their acceptor amino acids remains a major challenge. The attachment sites of the modification are difficult to localize by mass spectrometry (MS) because of the labile nature of the linkage and the complex fragmentation pattern of the ADP-ribose in MS/MS experiments. In this study we performed a comprehensive analysis of higher-energy collisional dissociation (HCD) spectra acquired from ADP-ribosylated peptides which were modified on arginine, serine, glutamic acid, aspartic acid, tyrosine, or lysine residues. In addition to the fragmentation of the peptide backbone, various cleavages of the ADP-ribosylated amino acid side chains were investigated. We focused on gas-phase fragmentations that were specific either to ADP-ribosylated arginine or to ADP-ribosylated serine and other O-linked ADP-ribosylations. The O-glycosidic linkage between ADP-ribose and serine, glutamic acid, or aspartic acid was the major cleavage site, making localization of these modification sites difficult. In contrast, the bond between ADP-ribose and arginine was relatively stable. The main cleavage site was the inner bond of the guanidine group, which resulted in the formation of ADP-ribosylated carbodiimide and of ornithine in place of modified arginine. Taking peptide fragment ions resulting from this specific cleavage into account, a considerably larger number of peptides containing ADP-ribosylated arginine were identified in database searches. Furthermore, the presence of diagnostic ions and of losses of fragments from peptide ions allowed us, in most cases, to distinguish between ADP-ribosylated arginine and serine residues.

Ultraviolet Photodissociation for Non-Target Screening-Based Identification of Organic Micro-Pollutants in Water Samples.

Non-target screening (NTS) based on the combination of liquid chromatography coupled to high-resolution mass spectrometry has become the key method to identify organic micro-pollutants (OMPs) in water samples. However, a large number of compounds remains unidentified with current NTS approaches due to poor quality fragmentation spectra generated by suboptimal fragmentation methods. Here, the potential of the alternative fragmentation technique ultraviolet photodissociation (UVPD) to improve identification of OMPs in water samples was investigated. A diverse set of water-relevant OMPs was selected based on k-means clustering and unsupervised artificial neural networks. The selected OMPs were analyzed using an Orbitrap Fusion Lumos equipped with UVPD. Therewith, information-rich MS2 fragmentation spectra of compounds that fragment poorly with higher-energy collisional dissociation (HCD) could be attained. Development of an R-based data analysis workflow and user interface facilitated the characterization and comparison of HCD and UVPD fragmentation patterns. UVPD and HCD generated both unique and common fragments, demonstrating that some fragmentation pathways are specific to the respective fragmentation method, while others seem more generic. Application of UVPD fragmentation to the analysis of surface water enabled OMP identification using existing HCD spectral libraries. However, high-throughput applications still require optimization of informatics workflows and spectral libraries tailored to UVPD.

Unraveling the proteomic changes involved in the resistance response of Cajanus platycarpus to herbivory by Helicoverpa armigera.

The pigeonpea wild relative Cajanus platycarpus is resistant to Helicoverpa armigera, one of the major pests responsible for yield losses in Cajanus cajan. Deciphering the molecular mechanism underlying host plant resistance is pertinent to identify proteins that aid in the mitigation of the insect pest. The present study adopted comparative proteomics as a tool to interpret the resistance mechanism(s) in C. platycarpus vis-à-vis C. cajan during continued herbivory (up to 96 h). Over-representation analysis of the differentially expressed proteins implicated a multi-dimensional resistance response accomplished by both physical and chemical barriers in C. platycarpus. While the chemical basis for resistance was depicted by the upregulation of proteins playing a rate limiting role in the phenylpropanoid pathway, the physical basis was provided by the regulation of proteins involved in microtubule assembly and synthesis of lignins. Upregulation of proteins in the polyamine pathway indicated the role of metabolite conjugates to be negatively affecting herbivore growth. Reallocation of resources and diversion of metabolic flux to support the production of secondary metabolites could be the probable approach in the wild relative against herbivory. Our study provided deeper insights into the pod borer resistance mechanism in C. platycarpus for utility in crop improvement. KEY POINTS: • Pod borer resistance in Cajanus platycarpus is multi-dimensional. • Pod borer resistance has been arbitrated to cell wall rigidity and secondary metabolites. • Phenylpropanoid pathway derivatives apparently shaped the plant chemical defense against pod borer.

Engineered peptide barcodes for in-depth analyses of binding protein libraries.

Binding protein generation typically relies on laborious screening cascades that process candidate molecules individually. We have developed NestLink, a binder selection and identification technology able to biophysically characterize thousands of library members at once without the need to handle individual clones at any stage of the process. NestLink uses genetically encoded barcoding peptides termed flycodes, which were designed for maximal detectability by mass spectrometry and support accurate deep sequencing. We demonstrate NestLink's capacity to overcome the current limitations of binder-generation methods in three applications. First, we show that hundreds of binder candidates can be simultaneously ranked according to kinetic parameters. Next, we demonstrate deep mining of a nanobody immune repertoire for membrane protein binders, carried out entirely in solution without target immobilization. Finally, we identify rare binders against an integral membrane protein directly in the cellular environment of a human pathogen. NestLink opens avenues for the selection of tailored binder characteristics directly in tissues or in living organisms.

In search of cerebrospinal fluid biomarkers of fatigue in multiple sclerosis: A proteomics study.

Fatigue in multiple sclerosis is a very common and cumbersome symptom, but its aetiology is poorly understood. Proteomics is increasingly implemented in multiple sclerosis research, but has not yet been used to study the neurobiological basis of fatigue in multiple sclerosis. To identify potential cerebrospinal fluid biomarkers of fatigue in multiple sclerosis, we collected cerebrospinal fluid of 20 patients with multiple sclerosis with fatigue (MS+), 20 patients with multiple sclerosis without fatigue (MS-), and 20 control subjects without multiple sclerosis and without fatigue (HC). We used a shotgun proteomics approach and label-free quantitative proteomics to analyse the protein content in cerebrospinal fluid. Selected proteins with differential abundance were further validated by immunoblotting. Out of 591 detected cerebrospinal fluid proteins, the abundance of nine proteins differed between the three groups, and seven additional proteins differed between MS+ and MS- patients. Using immunoblot or slot-blot techniques, we confirmed decreased levels of protein kinase C-binding protein NELL2, neural cell adhesion molecule L1-like protein, and reelin in MS+ patients. In conclusion, cerebrospinal fluid proteomics may provide insight into the neurobiological basis of fatigue in multiple sclerosis. The proteins identified to be decreased in MS+ are involved in synaptic plasticity and energy homeostasis, and thus appear as plausible biomarkers of this common symptom.

Gingival Exudatome Dynamics Implicate Inhibition of the Alternative Complement Pathway in the Protective Action of the C3 Inhibitor Cp40 in Nonhuman Primate Periodontitis.

Periodontitis is a prevalent chronic inflammatory disease associated with dysbiosis. Although complement inhibition has been successfully used to treat periodontitis in animal models, studies globally analyzing inflamed tissue proteins to glean insight into possible mechanisms of action are missing. Using quantitative shotgun proteomics, we aimed to investigate differences in composition of inflammatory gingival tissue exudate ("gingival crevicular fluid"; GCF), before and after local administration of an inhibitor of the central complement component, C3, in nonhuman primates. The C3 inhibitor, Cp40 (also known as AMY-101) was administered locally in the maxillary gingival tissue of cynomolgus monkeys with established periodontitis, either once a week (1×-treatment; n = 5 animals) or three times per week (3×-treatment; n = 10 animals), for 6 weeks followed by another 6 weeks of observation in the absence of treatment. 45 GCF samples were processed for FASP digestion and liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. Data were processed using the ProgenesisQI software. The statistical significance of differences between the groups was determined by RM-ANOVA, and a protein expression change was considered as a true regulation at >2-fold and p < 0.05. The human orthologues were subjected to Gene Ontology analyses using PANTHER. Data are available via ProteomeXchange with identifier PXD009502. 573 proteins with >2 peptides were longitudinally quantified. Both 3× and 1× administration of Cp40 resulted in significant down-regulation of dozens of proteins during the 6-week course of treatment as compared to baseline. Following drug withdrawal at 6 weeks, more than 50% of the down-regulated proteins showed increased levels at week 12. The top scored pathway was "complement activation, alternative pathway", and several proteins involved in this pathway were down-regulated at 6 weeks. We mapped the proteomic fingerprint changes in local tissue exudate of cynomolgus monkey periodontitis in response to C3 inhibition and identified the alternative pathway of complement activation and leukocyte degranulation as main targets, which are thus likely to play significant roles in periodontal disease pathogenesis. Label-free quantitative proteomics strategies utilizing GCF are powerful tools for the identification of treatment targets and providing insights into disease mechanisms.

rawDiag: An R Package Supporting Rational LC-MS Method Optimization for Bottom-up Proteomics.

Optimizing methods for liquid chromatography coupled to mass spectrometry (LC-MS) is a nontrivial task. Here we present rawDiag, a software tool supporting rational method optimization by providing MS operator-tailored diagnostic plots of scan-level metadata. rawDiag is implemented as an R package and can be executed on the R command line or through a graphical user interface (GUI) for less experienced users. The code runs platform-independent and can process 100 raw files in <3 min on current consumer hardware, as we show in our benchmark. As a demonstration of the functionality of our package we include a real-world example taken from our daily core facility business.

Combining Higher-Energy Collision Dissociation and Electron-Transfer/Higher-Energy Collision Dissociation Fragmentation in a Product-Dependent Manner Confidently Assigns Proteomewide ADP-Ribose Acceptor Sites.

Protein adenosine diphosphate (ADP)-ribosylation is a physiologically and pathologically important post-translational modification. Recent technological advances have improved analysis of this complex modification and have led to the discovery of hundreds of ADP-ribosylated proteins in both cultured cells and mouse tissues. Nevertheless, accurate assignment of the ADP-ribose acceptor site(s) within the modified proteins identified has remained a challenging task. This is mainly due to poor fragmentation of modified peptides. Here, using an Orbitrap Fusion Tribrid mass spectrometer, we present an optimized methodology that not only drastically improves the overall localization scores for ADP-ribosylation acceptor sites but also boosts ADP-ribosylated peptide identifications. First, we systematically compared the efficacy of higher-energy collision dissociation (HCD), electron-transfer dissociation with supplemental collisional activation (ETcaD), and electron-transfer/higher-energy collision dissociation (EThcD) fragmentation methods when determining ADP-ribose acceptor sites within complex cellular samples. We then tested the combination of HCD and EThcD fragmentation, which were employed in a product-dependent manner, and the unique fragmentation properties of the ADP-ribose moiety were used to trigger targeted fragmentation of only the modified peptides. The best results were obtained with a workflow that included initial fast, high-energy HCD (Orbitrap, FT) scans, which produced intense ADP-ribose fragmentation ions. These potentially ADP-ribosylated precursors were then selected and analyzed via subsequent high-resolution HCD and EThcD fragmentation. Using these resulting high-quality spectra, we identified a xxxxxxKSxxxxx modification motif where lysine can serve as an ADP-ribose acceptor site. Due to the appearance of serine within this motif and its close presence to the lysine, further analysis revealed that serine serves as a new ADP-ribose acceptor site across the proteome.

Targeted proteomics coming of age - SRM, PRM and DIA performance evaluated from a core facility perspective.

Quantitative mass spectrometry is a rapidly evolving methodology applied in a large number of omics-type research projects. During the past years, new designs of mass spectrometers have been developed and launched as commercial systems while in parallel new data acquisition schemes and data analysis paradigms have been introduced. Core facilities provide access to such technologies, but also actively support the researchers in finding and applying the best-suited analytical approach. In order to implement a solid fundament for this decision making process, core facilities need to constantly compare and benchmark the various approaches. In this article we compare the quantitative accuracy and precision of current state of the art targeted proteomics approaches single reaction monitoring (SRM), parallel reaction monitoring (PRM) and data independent acquisition (DIA) across multiple liquid chromatography mass spectrometry (LC-MS) platforms, using a readily available commercial standard sample. All workflows are able to reproducibly generate accurate quantitative data. However, SRM and PRM workflows show higher accuracy and precision compared to DIA approaches, especially when analyzing low concentrated analytes.

specL--an R/Bioconductor package to prepare peptide spectrum matches for use in targeted proteomics.

MOTIVATION: Targeted data extraction methods are attractive ways to obtain quantitative peptide information from a proteomics experiment. Sequential Window Acquisition of all Theoretical Spectra (SWATH) and Data Independent Acquisition (DIA) methods increase reproducibility of acquired data because the classical precursor selection is omitted and all present precursors are fragmented. However, especially for targeted data extraction, MS coordinates (retention time information precursor and fragment masses) are required for the particular entities (peptide ions). These coordinates are usually generated in a so-called discovery experiment earlier on in the project if not available in public spectral library repositories. The quality of the assay panel is crucial to ensure appropriate downstream analysis. For that, a method is needed to create spectral libraries and to export customizable assay panels. RESULTS: Here, we present a versatile set of functions to generate assay panels from spectral libraries for use in targeted data extraction methods (SWATH/DIA) in the area of proteomics. AVAILABILITY AND IMPLEMENTATION: specL is implemented in the R language and available under an open-source license (GPL-3) in Bioconductor since BioC 3.0 (R-3.1) http://www.bioconductor.org (Trachsel et al., 2015). A vignette with a complete tutorial describing data import/export and analysis is included in the package and can also be found as supplement material of this article. CONTACT: cp@fgcz.ethz.ch or jg@fgcz.ethz.ch SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

PTM MarkerFinder, a software tool to detect and validate spectra from peptides carrying post-translational modifications.

Mass spectrometry (MS) analysis of peptides carrying post-translational modifications is challenging due to the instability of some modifications during MS analysis. However, glycopeptides as well as acetylated, methylated and other modified peptides release specific fragment ions during CID (collision-induced dissociation) and HCD (higher energy collisional dissociation) fragmentation. These fragment ions can be used to validate the presence of the PTM on the peptide. Here, we present PTM MarkerFinder, a software tool that takes advantage of such marker ions. PTM MarkerFinder screens the MS/MS spectra in the output of a database search (i.e., Mascot) for marker ions specific for selected PTMs. Moreover, it reports and annotates the HCD and the corresponding electron transfer dissociation (ETD) spectrum (when present), and summarizes information on the type, number, and ratios of marker ions found in the data set. In the present work, a sample containing enriched N-acetylhexosamine (HexNAc) glycopeptides from yeast has been analyzed by liquid chromatography-mass spectrometry on an LTQ Orbitrap Velos using both HCD and ETD fragmentation techniques. The identification result (Mascot .dat file) was submitted as input to PTM MarkerFinder and screened for HexNAc oxonium ions. The software output has been used for high-throughput validation of the identification results.

FCC - An automated rule-based processing tool for life science data.

BACKGROUND: Data processing in the bioinformatics field often involves the handling of diverse software programs in one workflow. The field is lacking a set of standards for file formats so that files have to be processed in different ways in order to make them compatible to different analysis programs. The problem is that mass spectrometry vendors at most provide only closed-source Windows libraries to programmatically access their proprietary binary formats. This prohibits the creation of an efficient and unified tool that fits all processing needs of the users. Therefore, researchers are spending a significant amount of time using GUI-based conversion and processing programs. Besides the time needed for manual usage, such programs also can show long running times for processing, because most of them make use of only a single CPU. In particular, algorithms to enhance data quality, e.g. peak picking or deconvolution of spectra, add waiting time for the users. RESULTS: To automate these processing tasks and let them run continuously without user interaction, we developed the FGCZ Converter Control (FCC) at the Functional Genomics Center Zurich (FGCZ) core facility. The FCC is a rule-based system for automated file processing that reduces the operation of diverse programs to a single configuration task. Using filtering rules for raw data files, the parameters for all tasks can be custom-tailored to the needs of every single researcher and processing can run automatically and efficiently on any number of servers in parallel using all available CPU resources. CONCLUSIONS: FCC has been used intensively at FGCZ for processing more than hundred thousand mass spectrometry raw files so far. Since we know that many other research facilities have similar problems, we would like to report on our tool and the accompanying ideas for an efficient set-up for potential reuse.

Identification of combinatorial patterns of post-translational modifications on individual histones in the mouse brain.

Post-translational modifications (PTMs) of proteins are biochemical processes required for cellular functions and signalling that occur in every sub-cellular compartment. Multiple protein PTMs exist, and are established by specific enzymes that can act in basal conditions and upon cellular activity. In the nucleus, histone proteins are subjected to numerous PTMs that together form a histone code that contributes to regulate transcriptional activity and gene expression. Despite their importance however, histone PTMs have remained poorly characterised in most tissues, in particular the brain where they are thought to be required for complex functions such as learning and memory formation. Here, we report the comprehensive identification of histone PTMs, of their combinatorial patterns, and of the rules that govern these patterns in the adult mouse brain. Based on liquid chromatography, electron transfer, and collision-induced dissociation mass spectrometry, we generated a dataset containing a total of 10,646 peptides from H1, H2A, H2B, H3, H4, and variants in the adult brain. 1475 of these peptides carried one or more PTMs, including 141 unique sites and a total of 58 novel sites not described before. We observed that these PTMs are not only classical modifications such as serine/threonine (Ser/Thr) phosphorylation, lysine (Lys) acetylation, and Lys/arginine (Arg) methylation, but also include several atypical modifications such as Ser/Thr acetylation, and Lys butyrylation, crotonylation, and propionylation. Using synthetic peptides, we validated the presence of these atypical novel PTMs in the mouse brain. The application of data-mining algorithms further revealed that histone PTMs occur in specific combinations with different ratios. Overall, the present data newly identify a specific histone code in the mouse brain and reveal its level of complexity, suggesting its potential relevance for higher-order brain functions.