Protein Inference Research Articles

Improved Protein Inference from Multiple Protease Bottom-Up Mass Spectrometry Data.

Peptides detected by tandem mass spectrometry (MS/MS) in bottom-up proteomics serve as proxies for the proteins expressed in the sample. Protein inference is a process routinely applied to these peptides to generate a plausible list of candidate protein identifications. The use of multiple proteases for parallel protein digestions expands sequence coverage, provides additional peptide identifications, and increases the probability of identifying peptides that are unique to a single protein, which are all valuable for protein inference. We have developed and implemented a multi-protease protein inference algorithm in MetaMorpheus, a bottom-up search software program, which incorporates the calculation of protease-specific q-values and preserves the association of peptide sequences and their protease of origin. This integrated multi-protease protein inference algorithm provides more accurate results than either the aggregation of results from the separate analysis of the peptide identifications produced by each protease (separate approach) in MetaMorpheus, or results that are obtained using Fido, ProteinProphet, or DTASelect2. MetaMorpheus' integrated multi-protease data analysis decreases the ambiguity of the protein group list, reduces the frequency of erroneous identifications, and increases the number of post-translational modifications identified, while combining multi-protease search and protein inference into a single software program.

Statistical compression of protein sequences and inference of marginal probability landscapes over competing alignments using finite state models and Dirichlet priors.

The information criterion of minimum message length (MML) provides a powerful statistical framework for inductive reasoning from observed data. We apply MML to the problem of protein sequence comparison using finite state models with Dirichlet distributions. The resulting framework allows us to supersede the ad hoc cost functions commonly used in the field, by systematically addressing the problem of arbitrariness in alignment parameters, and the disconnect between substitution scores and gap costs. Furthermore, our framework enables the generation of marginal probability landscapes over all possible alignment hypotheses, with potential to facilitate the users to simultaneously rationalize and assess competing alignment relationships between protein sequences, beyond simply reporting a single (best) alignment. We demonstrate the performance of our program on benchmarks containing distantly related protein sequences.Availability and implementationThe open-source program supporting this work is available from: http://lcb.infotech.monash.edu.au/seqmmligner.Supplementary information Supplementary data are available at Bioinformatics online.

DESTINI: A deep-learning approach to contact-driven protein structure prediction

The amino acid sequence of a protein encodes the blueprint of its native structure. To predict the corresponding structural fold from the protein’s sequence is one of most challenging problems in computational biology. In this work, we introduce DESTINI (deep structural inference for proteins), a novel computational approach that combines a deep-learning algorithm for protein residue/residue contact prediction with template-based structural modelling. For the first time, the significantly improved predictive ability is demonstrated in the large-scale tertiary structure prediction of over 1,200 single-domain proteins. DESTINI successfully predicts the tertiary structure of four times the number of “hard” targets (those with poor quality templates) that were previously intractable, viz, a “glass-ceiling” for previous template-based approaches, and also improves model quality for “easy” targets (those with good quality templates). The significantly better performance by DESTINI is largely due to the incorporation of better contact prediction into template modelling. To understand why deep-learning accomplishes more accurate contact prediction, systematic clustering reveals that deep-learning predicts coherent, native-like contact patterns compared to co-evolutionary analysis. Taken together, this work presents a promising strategy towards solving the protein structure prediction problem.

Computational Approaches for Understanding High Quality Mass Spectrometry Proteomic Data

Computational approach of proteomic data science is used to identification and quantification of protein and provides the high-throughput data, concentration changes, interactions, posttranslational modifications and cellular localizations. The high-quality mass spectrometry recall to understanding the different sources of unsigned high-quality spectra features. The iterative computational method is interrogating the high efficiency of mass spectrometry protein data. The approach contains several databases searching with different search parameters, spectral library searching, modified peptides using blind search and genomic database searching. The mass spectrometry computational method is analysis the proteomics data focusing the key concepts with explanations, mass spectral feature detection, identifying the peptides, protein inference and control the false discovery rate. Then the method discusses the quantification of peptides and proteins, the downstream data analysis on machine learning, network analysis and multiomics integration of protein data and finally discuss the future of computational proteomics data.

Improved inference of site-specific positive selection under a generalized parametric codon model when there are multinucleotide mutations and multiple nonsynonymous rates

BackgroundAn excess of nonsynonymous substitutions, over neutrality, is considered evidence of positive Darwinian selection. Inference for proteins often relies on estimation of the nonsynonymous to synonymous ratio (ω = dN/dS) within a codon model. However, to ease computational difficulties, ω is typically estimated assuming an idealized substitution process where (i) all nonsynonymous substitutions have the same rate (regardless of impact on organism fitness) and (ii) instantaneous double and triple (DT) nucleotide mutations have zero probability (despite evidence that they can occur). It follows that estimates of ω represent an imperfect summary of the intensity of selection, and that tests based on the ω > 1 threshold could be negatively impacted.ResultsWe developed a general-purpose parametric (GPP) modelling framework for codons. This novel approach allows specification of all possible instantaneous codon substitutions, including multiple nonsynonymous rates (MNRs) and instantaneous DT nucleotide changes. Existing codon models are specified as special cases of the GPP model. We use GPP models to implement likelihood ratio tests for ω > 1 that accommodate MNRs and DT mutations. Through both simulation and real data analysis, we find that failure to model MNRs and DT mutations reduces power in some cases and inflates false positives in others. False positives under traditional M2a and M8 models were very sensitive to DT changes. This was exacerbated by the choice of frequency parameterization (GY vs. MG), with rates sometimes > 90% under MG. By including MNRs and DT mutations, accuracy and power was greatly improved under the GPP framework. However, we also find that over-parameterized models can perform less well, and this can contribute to degraded performance of LRTs.ConclusionsWe suggest GPP models should be used alongside traditional codon models. Further, all codon models should be deployed within an experimental design that includes (i) assessing robustness to model assumptions, and (ii) investigation of non-standard behaviour of MLEs. As the goal of every analysis is to avoid false conclusions, more work is needed on model selection methods that consider both the increase in fit engendered by a model parameter and the degree to which that parameter is affected by un-modelled evolutionary processes.

Middle-down approach: a choice to sequence and characterize proteins/proteomes by mass spectrometry.

Owing to rapid growth in the elucidation of genome sequences of various organisms, deducing proteome sequences has become imperative, in order to have an improved understanding of biological processes. Since the traditional Edman method was unsuitable for high-throughput sequencing and also for N-terminus modified proteins, mass spectrometry (MS) based methods, mainly based on soft ionization modes: electrospray ionization and matrix-assisted laser desorption/ionization, began to gain significance. MS based methods were adaptable for high-throughput studies and applicable for sequencing N-terminus blocked proteins/peptides too. Consequently, over the last decade a new discipline called ‘proteomics’ has emerged, which encompasses the attributes necessary for high-throughput identification of proteins. ‘Proteomics’ may also be regarded as an offshoot of the classic field, ‘biochemistry’. Many protein sequencing and proteomic investigations were successfully accomplished through MS dependent sequence elucidation of ‘short proteolytic peptides (typically: 7–20 amino acid residues), which is called the ‘shotgun’ or ‘bottom-up (BU)’ approach. While the BU approach continues as a workhorse for proteomics/protein sequencing, attempts to sequence intact proteins without proteolysis, called the ‘top-down (TD)’ approach started, due to ambiguities in the BU approach, e.g., protein inference problem, identification of proteoforms and the discovery of posttranslational modifications (PTMs). The high-throughput TD approach (TD proteomics) is yet in its infancy. Nevertheless, TD characterization of purified intact proteins has been useful for detecting PTMs. With the hope to overcome the pitfalls of BU and TD strategies, another concept called the ‘middle-down (MD)’ approach was put forward. Similar to BU, the MD approach also involves proteolysis, but in a restricted manner, to produce ‘longer’ proteolytic peptides than the ones usually obtained in BU studies, thereby providing better sequence coverage. In this regard, special proteases (OmpT, Sap9, IdeS) have been used, which can cleave proteins to produce longer proteolytic peptides. By reviewing ample evidences currently existing in the literature that is predominantly on PTM characterization of histones and antibodies, herein we highlight salient features of the MD approach. Consequently, we are inclined to claim that the MD concept might have widespread applications in future for various research areas, such as clinical, biopharmaceuticals (including PTM analysis) and even for general/routine characterization of proteins including therapeutic proteins, but not just limited to analysis of histones or antibodies.

A comprehensive pipeline for translational top-down proteomics from a single blood draw.

Top-down proteomics (TDP) by mass spectrometry (MS) is a technique by which intact proteins are analyzed. It has become increasingly popDesalting and concentrating GELFrEEular in translational research because of the value of characterizing distinct proteoforms of intact proteins. Compared to bottom-up proteomics (BUP) strategies, which measure digested peptide mixtures, TDP provides highly specific molecular information that avoids the bioinformatic challenge of protein inference. However, the technique has been difficult to implement widely because of inherent limitations of existing sample preparation methods and instrumentation. Recent improvements in proteoform pre-fractionation and the availability of high-resolution benchtop mass spectrometers have made it possible to use high-throughput TDP for the analysis of complex clinical samples. Here, we provide a comprehensive protocol for analysis of a common sample type in translational research: human peripheral blood mononuclear cells (PBMCs). The pipeline comprises multiple workflows that can be treated as modular by the reader and used for various applications. First, sample collection and cell preservation are described for two clinical biorepository storage schemes. Cell lysis and proteoform pre-fractionation by gel-eluted liquid fractionation entrapment electrophoresis are then described. Importantly, instrument setup and liquid chromatography-tandem MS are described for TDP analyses, which rely on high-resolution Fourier-transform MS. Finally, data processing and analysis are described using two different, application-dependent software tools: ProSight Lite for targeted analyses of one or a few proteoforms and TDPortal for high-throughput TDP in discovery mode. For a single sample, the minimum completion time of the entire experiment is 72 h.

Protein Inference Using PIA Workflows and PSI Standard File Formats.

Proteomics using LC-MS/MS has become one of the main methods to analyze the proteins in biological samples in high-throughput. But the existing mass-spectrometry instruments are still limited with respect to resolution and measurable mass ranges, which is one of the main reasons why shotgun proteomics is the major approach. Here proteins are digested, which leads to the identification and quantification of peptides instead. While often neglected, the important step of protein inference needs to be conducted to infer from the identified peptides to the actual proteins in the original sample. In this work, we highlight some of the previously published and newly added features of the tool PIA - Protein Inference Algorithms, which helps the user with the protein inference of measured samples. We also highlight the importance of the usage of PSI standard file formats, as PIA is the only current software supporting all available standards used for spectrum identification and protein inference. Additionally, we briefly describe the benefits of working with workflow environments for proteomics analyses and show the new features of the PIA nodes for the KNIME Analytics Platform. Finally, we benchmark PIA against a recently published data set for isoform detection. PIA is open source and available for download on GitHub ( https://github.com/mpc-bioinformatics/pia ) or directly via the community extensions inside the KNIME analytics platform.

A perspective view of top-down proteomics in snake venom research.

The venom produced by snakes contains complex mixtures of pharmacologically active proteins and peptides which play a crucial role in the pathophysiology of snakebite diseases. The deep understanding of venom proteomes can help to improve the treatment of this "neglected tropical disease" (as expressed by the World Health Organization [WHO]) and to develop new drugs. The most widely used technique for venom analysis is liquid chromatography/tandem mass spectrometry (LC/MS/MS)-based bottom-up (BU) proteomics. Considering the fact that multiple multi-locus gene families encode snake venom proteins, the major challenge for the BU proteomics is the limited sequence coverage and also the "protein inference problem" which result in a loss of information for the identification and characterization of toxin proteoforms (genetic variation, alternative mRNA splicing, single nucleotide polymorphism [SNP] and post-translational modifications [PTMs]). In contrast, intact protein measurements with top-down (TD) MS strategies cover almost complete protein sequences, and prove the ability to identify venom proteoforms and to localize their modifications and sequence variations.

Computational Methods for Understanding Mass Spectrometry–Based Shotgun Proteomics Data

Computational proteomics is the data science concerned with the identification and quantification of proteins from high-throughput data and the biological interpretation of their concentration changes, posttranslational modifications, interactions, and subcellular localizations. Today, these data most often originate from mass spectrometry–based shotgun proteomics experiments. In this review, we survey computational methods for the analysis of such proteomics data, focusing on the explanation of the key concepts. Starting with mass spectrometric feature detection, we then cover methods for the identification of peptides. Subsequently, protein inference and the control of false discovery rates are highly important topics covered. We then discuss methods for the quantification of peptides and proteins. A section on downstream data analysis covers exploratory statistics, network analysis, machine learning, and multiomics data integration. Finally, we discuss current developments and provide an outlook on what the near future of computational proteomics might bear.

Transcription factor activating protein 4 is synthetically lethal and a master regulator of MYCN-amplified neuroblastoma

Despite the identification of MYCN amplification as an adverse prognostic marker in neuroblastoma, MYCN inhibitors have yet to be developed. Here, by integrating evidence from a whole-genome shRNA library screen and the computational inference of master regulator proteins, we identify transcription factor activating protein 4 (TFAP4) as a critical effector of MYCN amplification in neuroblastoma, providing a novel synthetic lethal target. We demonstrate that TFAP4 is a direct target of MYCN in neuroblastoma cells, and that its expression and activity strongly negatively correlate with neuroblastoma patient survival. Silencing TFAP4 selectively inhibits MYCN-amplified neuroblastoma cell growth both in vitro and in vivo, in xenograft mouse models. Mechanistically, silencing TFAP4 induces neuroblastoma differentiation, as evidenced by increased neurite outgrowth and upregulation of neuronal markers. Taken together, our results demonstrate that TFAP4 is a key regulator of MYCN-amplified neuroblastoma and may represent a valuable novel therapeutic target.

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.

A Protein Standard That Emulates Homology for the Characterization of Protein Inference Algorithms.

A natural way to benchmark the performance of an analytical experimental setup is to use samples of known composition and see to what degree one can correctly infer the content of such a sample from the data. For shotgun proteomics, one of the inherent problems of interpreting data is that the measured analytes are peptides and not the actual proteins themselves. As some proteins share proteolytic peptides, there might be more than one possible causative set of proteins resulting in a given set of peptides and there is a need for mechanisms that infer proteins from lists of detected peptides. A weakness of commercially available samples of known content is that they consist of proteins that are deliberately selected for producing tryptic peptides that are unique to a single protein. Unfortunately, such samples do not expose any complications in protein inference. Hence, for a realistic benchmark of protein inference procedures, there is a need for samples of known content where the present proteins share peptides with known absent proteins. Here, we present such a standard, that is based on E. coli expressed human protein fragments. To illustrate the application of this standard, we benchmark a set of different protein inference procedures on the data. We observe that inference procedures excluding shared peptides provide more accurate estimates of errors compared to methods that include information from shared peptides, while still giving a reasonable performance in terms of the number of identified proteins. We also demonstrate that using a sample of known protein content without proteins with shared tryptic peptides can give a false sense of accuracy for many protein inference methods.

DeepPep: Deep proteome inference from peptide profiles.

Protein inference, the identification of the protein set that is the origin of a given peptide profile, is a fundamental challenge in proteomics. We present DeepPep, a deep-convolutional neural network framework that predicts the protein set from a proteomics mixture, given the sequence universe of possible proteins and a target peptide profile. In its core, DeepPep quantifies the change in probabilistic score of peptide-spectrum matches in the presence or absence of a specific protein, hence selecting as candidate proteins with the largest impact to the peptide profile. Application of the method across datasets argues for its competitive predictive ability (AUC of 0.80±0.18, AUPR of 0.84±0.28) in inferring proteins without need of peptide detectability on which the most competitive methods rely. We find that the convolutional neural network architecture outperforms the traditional artificial neural network architectures without convolution layers in protein inference. We expect that similar deep learning architectures that allow learning nonlinear patterns can be further extended to problems in metagenome profiling and cell type inference. The source code of DeepPep and the benchmark datasets used in this study are available at https://deeppep.github.io/DeepPep/.

Model for Quality Control of Allergen Products with Mass Spectrometry.

Birch pollen allergy is diagnosed and treated with aqueous extracts from birch pollen, which contain a mixture of allergens and nonallergenic proteins, including large numbers of closely related sequence variants, so-called iso-allergens of the major allergen, Bet v 1. The quality of therapeutic and diagnostic allergen products largely depends on the allergen and iso-allergen composition. Several biochemical methods are currently applied to detect and quantify allergens and to record protein profiles without differentiating between iso-allergens. Mass spectrometry (MS) may entirely replace these technologies, as it allows sequence specific identification and quantification of proteins and protein profiles including sequence variants in one run. However, the protein inference problem still hampers the automatic assignment of peptide sequences to proteins, consequently impeding the quantification of sequence variants. Therefore, the aim of the study was to set up semitargeted analyses of label-free MS data that allow unambiguous identification and quantification of birch pollen allergens and nonallergenic proteins. We combined data independent acquisition with manual assignment of predefined target sequences for quantification of iso-allergens and automatic quantification of other allergens and nonallergenic proteins. The quantitative data for birch pollen allergens and sequence variants of Bet v 1 were further confirmed by multiple reaction monitoring.

High-throughput metaproteomics data analysis with Unipept: A tutorial

In recent years, shotgun metaproteomics has established itself as an important tool to study the composition of complex ecosystems and microbial communities. Two key steps in metaproteomics data analysis are the inference of proteins from the identified peptides, and the determination of the taxonomic origin and function of these proteins. This tutorial therefore introduces the Unipept command line interface (http://unipept.ugent.be/clidocs) as a platform-independent tool for such metaproteomics data analyses. First, a detailed overview is given of the available Unipept commands and their functions. Next, the power of the Unipept command line interface is illustrated using two case studies that analyze a single tryptic peptide, and a set of peptides retrieved from a shotgun metaproteomics experiment, respectively. Finally, the analysis results obtained using these command line tools are compared with the interactive taxonomic analysis that is available on the Unipept website.

The mzIdentML Data Standard Version 1.2, Supporting Advances in Proteome Informatics.

The first stable version of the Proteomics Standards Initiative mzIdentML open data standard (version 1.1) was published in 2012—capturing the outputs of peptide and protein identification software. In the intervening years, the standard has become well-supported in both commercial and open software, as well as a submission and download format for public repositories. Here we report a new release of mzIdentML (version 1.2) that is required to keep pace with emerging practice in proteome informatics. New features have been added to support: (1) scores associated with localization of modifications on peptides; (2) statistics performed at the level of peptides; (3) identification of cross-linked peptides; and (4) support for proteogenomics approaches. In addition, there is now improved support for the encoding of de novo sequencing of peptides, spectral library searches, and protein inference. As a key point, the underlying XML schema has only undergone very minor modifications to simplify as much as possible the transition from version 1.1 to version 1.2 for implementers, but there have been several notable updates to the format specification, implementation guidelines, controlled vocabularies and validation software. mzIdentML 1.2 can be described as backwards compatible, in that reading software designed for mzIdentML 1.1 should function in most cases without adaptation. We anticipate that these developments will provide a continued stable base for software teams working to implement the standard. All the related documentation is accessible at http://www.psidev.info/mzidentml.

Predicting genome terminus sequences of Bacillus cereus-group bacteriophage using next generation sequencing data

BackgroundMost tailed bacteriophages (phages) feature linear dsDNA genomes. Characterizing novel phages requires an understanding of complete genome sequences, including the definition of genome physical ends.ResultWe sequenced 48 Bacillus cereus phage isolates and analyzed Next-generation sequencing (NGS) data to resolve the genome configuration of these novel phages. Most assembled contigs featured reads that mapped to both contig ends and formed circularized contigs. Independent assemblies of 31 nearly identical I48-like Bacillus phage isolates allowed us to observe that the assembly programs tended to produce random cleavage on circularized contigs. However, currently available assemblers were not capable of reporting the underlying phage genome configuration from sequence data. To identify the genome configuration of sequenced phage in silico, a terminus prediction method was developed by means of ‘neighboring coverage ratios’ and ‘read edge frequencies’ from read alignment files. Termini were confirmed by primer walking and supported by phylogenetic inference of large DNA terminase protein sequences.ConclusionsThe Terminus package using phage NGS data along with the contig circularity could efficiently identify the proximal positions of phage genome terminus. Complete phage genome sequences allow a proposed characterization of the potential packaging mechanisms and more precise genome annotation.

Combinatorial Labeling Method for Improving Peptide Fragmentation in Mass Spectrometry.

Annotation of peptide sequence from tandem mass spectra constitutes the central step of mass spectrometry-based proteomics. Peptide mass spectra are obtained upon gas-phase fragmentation. Identification of the protein from a set of experimental peptide spectral matches is usually referred as protein inference. Occurrence and intensity of these fragment ions in the MS/MS spectra are dependent on many factors such as amino acid composition, peptide basicity, activation mode, protease, etc. Particularly, chemical derivatizations of peptides were known to alter their fragmentation. In this study, the influence of acetylation, guanidinylation, and their combination on peptide fragmentation was assessed initially on a lipase (LipA) from Bacillus subtilis followed by a bovine six protein mix digest. The dual modification resulted in improved fragment ion occurrence and intensity changes, and this resulted in the equivalent representation of b- and y-type fragment ions in an ion trap MS/MS spectrum. The improved representation has allowed us to accurately annotate the peptide sequences de novo. Dual labeling has significantly reduced the false positive protein identifications in standard bovine six peptide digest. Our study suggests that the combinatorial labeling of peptides is a useful method to validate protein identifications for high confidence protein inference. Graphical Abstract ᅟ.

X!TandemPipeline: A Tool to Manage Sequence Redundancy for Protein Inference and Phosphosite Identification.

X!TandemPipeline is a software designed to perform protein inference and to manage redundancy in the results of phosphosite identification by database search. It provides the minimal list of proteins or phosphosites that are present in a set of samples using grouping algorithms based on the principle of parsimony. Regarding proteins, a two-level classification is performed, where groups gather proteins sharing at least one peptide and subgroups gather proteins that are not distinguishable according to the identified peptides. Regarding phosphosites, an innovative approach based on the concept of phosphoisland is used to gather overlapping phosphopeptides. The graphical interface of X!TandemPipeline allows the users to launch X!tandem identification, to inspect spectra and to manually validate their assignment to peptides, to launch the grouping program, and to visualize elementary data as well as grouping and redundancy information. Identification results obtained from other search engines can also be processed. X!TandemPipeline results can be exported as ready-to-use tabulated files or as XML files that can be directly used by the PROTICdb database or by the MassChroQ quantification software. X!TandemPipeline runs fast, is easy to use, and can process hundreds of samples simultaneously. It is freely available under the GNU General Public License v3.0 at http://pappso.inra.fr/bioinfo/xtandempipeline/ .