A note on the false discovery rate of novel peptides in proteogenomics

Small proteins in bacteria - Big challenges in prediction and identification

Proteins with up to 100 amino acids have been largely overlooked due to the challenges associated with predicting and identifying them using traditional methods. Recent advances in bioinformatics and machine learning, DNA sequencing, RNA and Ribo-seq technologies, and mass spectrometry (MS) have greatly facilitated the detection and characterisation of these elusive proteins in recent years. This has revealed their crucial role in various cellular processes including regulation, signalling and transport, as toxins and as folding helpers for protein complexes. Consequently, the systematic identification and characterisation of these proteins in bacteria have emerged as a prominent field of interest within the microbial research community. This review provides an overview of different strategies for predicting and identifying these proteins on a large scale, leveraging the power of these advanced technologies. Furthermore, the review offers insights into the future developments that may be expected in this field.

Microproteins-Discovery, structure, and function

Advances in proteogenomic technologies have revealed hundreds to thousands of translated small open reading frames (sORFs) that encode microproteins in genomes across evolutionary space. While many microproteins have now been shown to play critical roles in biology and human disease, a majority of recently identified microproteins have little or no experimental evidence regarding their functionality. Computational tools have some limitations for analysis of short, poorly conserved microprotein sequences, so additional approaches are needed to determine the role of each member of this recently discovered polypeptide class. A currently underexplored avenue in the study of microproteins is structure prediction and determination, which delivers a depth of functional information. In this review, we provide a brief overview of microprotein discovery methods, then examine examples of microprotein structures (and, conversely, intrinsic disorder) that have been experimentally determined using crystallography, cryo-electron microscopy, and NMR, which provide insight into their molecular functions and mechanisms. Additionally, we discuss examples of predicted microprotein structures that have provided insight or context regarding their function. Analysis of microprotein structure at the angstrom level, and confirmation of predicted structures, therefore, has potential to identify translated microproteins that are of biological importance and to provide molecular mechanism for their in vivo roles.

Chemical labeling and proteomics for characterization of unannotated small and alternative open reading frame-encoded polypeptides

Thousands of unannotated small and alternative open reading frames (smORFs and alt-ORFs, respectively) have recently been revealed in mammalian genomes. While hundreds of mammalian smORF- and alt-ORF-encoded proteins (SEPs and alt-proteins, respectively) affect cell proliferation, the overwhelming majority of smORFs and alt-ORFs remain uncharacterized at the molecular level. Complicating the task of identifying the biological roles of smORFs and alt-ORFs, the SEPs and alt-proteins that they encode exhibit limited sequence homology to protein domains of known function. Experimental techniques for the functionalization of these gene classes are therefore required. Approaches combining chemical labeling and quantitative proteomics have greatly advanced our ability to identify and characterize functional SEPs and alt-proteins in high throughput. In this review, we briefly describe the principles of proteomic discovery of SEPs and alt-proteins, then summarize how these technologies interface with chemical labeling for identification of SEPs and alt-proteins with specific properties, as well as in defining the interactome of SEPs and alt-proteins.

Mirror proteases of Ac-Trypsin and Ac-LysargiNase precisely improve novel event identifications in Mycolicibacterium smegmatis MC2 155 by proteogenomic analysis

Accurate identification of novel peptides remains challenging because of the lack of evaluation criteria in large-scale proteogenomic studies. Mirror proteases of trypsin and lysargiNase can generate complementary b/y ion series, providing the opportunity to efficiently assess authentic novel peptides in experiments other than filter potential targets by different false discovery rates (FDRs) ranking. In this study, a pair of in-house developed acetylated mirror proteases, Ac-Trypsin and Ac-LysargiNase, were used in Mycolicibacterium smegmatis MC² 155 for proteogenomic analysis. The mirror proteases accurately identified 368 novel peptides, exhibiting 75–80% b and y ion coverages against 65–68% y or b ion coverages of Ac-Trypsin (38.9% b and 68.3% y) or Ac-LysargiNase (65.5% b and 39.6% y) as annotated peptides from M. smegmatis MC² 155. The complementary b and y ion series largely increased the reliability of overlapped sequences derived from novel peptides. Among these novel peptides, 311 peptides were annotated in other public M. smegmatis strains, and 57 novel peptides with more continuous b and y pairs were obtained for further analysis after spectral quality assessment. This enabled mirror proteases to successfully correct six annotated proteins' N-termini and detect 17 new coding open reading frames (ORFs). We believe that mirror proteases will be an effective strategy for novel peptide detection in both prokaryotic and eukaryotic proteogenomics.

False discovery rate: the Achilles' heel of proteogenomics

Proteogenomics refers to the integrated analysis of the genome and proteome that leverages mass-spectrometry (MS)-based proteomics data to improve genome annotations, understand gene expression control through proteoforms and find sequence variants to develop novel insights for disease classification and therapeutic strategies. However, proteogenomic studies often suffer from reduced sensitivity and specificity due to inflated database size. To control the error rates, proteogenomics depends on the target-decoy search strategy, the de-facto method for false discovery rate (FDR) estimation in proteomics. The proteogenomic databases constructed from three- or six-frame nucleotide database translation not only increase the search space and compute-time but also violate the equivalence of target and decoy databases. These searches result in poorer separation between target and decoy scores, leading to stringent FDR thresholds. Understanding these factors and applying modified strategies such as two-pass database search or peptide-class-specific FDR can result in a better interpretation of MS data without introducing additional statistical biases. Based on these considerations, a user can interpret the proteogenomics results appropriately and control false positives and negatives in a more informed manner. In this review, first, we briefly discuss the proteogenomic workflows and limitations in database construction, followed by various considerations that can influence potential novel discoveries in a proteogenomic study. We conclude with suggestions to counter these challenges for better proteogenomic data interpretation.

Bottom-up and top-down proteomic approaches for the identification, characterization, and quantification of the low molecular weight proteome with focus on short open reading frame-encoded peptides

The recent discovery of alternative open reading frames creates a need for suitable analytical approaches to verify their translation and to characterize the corresponding gene products at the molecular level. As the analysis of small proteins within a background proteome by means of classical bottom-up proteomics is challenging, method development for the analysis of small open reading frame encoded peptides (SEPs) have become a focal point for research. Here, we highlight bottom-up and top-down proteomics approaches established for the analysis of SEPs in both pro- and eukaryotes. Major steps of analysis, including sample preparation and (small) proteome isolation, separation and mass spectrometry, data interpretation and quality control, quantification, the analysis of post-translational modifications, and exploration of functional aspects of the SEPs by means of proteomics technologies are described. These methods do not exclusively cover the analytics of SEPs but simultaneously include the low molecular weight proteome, and moreover, can also be used for the proteome-wide analysis of proteolytic processing events.

A workflow to identify novel proteins based on the direct mapping of peptide-spectrum-matches to genomic locations

Background

Small Proteins have received increasing attention in recent years. They have in particular been implicated as signals contributing to the coordination of bacterial communities. In genome annotations they are often missing or hidden among large numbers of hypothetical proteins because genome annotation pipelines often exclude short open reading frames or over-predict hypothetical proteins based on simple models. The validation of novel proteins, and in particular of small proteins (sProteins), therefore requires additional evidence. Proteogenomics is considered the gold standard for this purpose. It extends beyond established annotations and includes all possible open reading frames (ORFs) as potential sources of peptides, thus allowing the discovery of novel, unannotated proteins. Typically this results in large numbers of putative novel small proteins fraught with large fractions of false-positive predictions.

Results

We observe that number and quality of the peptide-spectrum matches (PSMs) that map to a candidate ORF can be highly informative for the purpose of distinguishing proteins from spurious ORF annotations. We report here on a workflow that aggregates PSM quality information and local context into simple descriptors and reliably separates likely proteins from the large pool of false-positive, i.e., most likely untranslated ORFs. We investigated the artificial gut microbiome model SIHUMIx, comprising eight different species, for which we validate 5114 proteins that have previously been annotated only as hypothetical ORFs. In addition, we identified 37 non-annotated protein candidates for which we found evidence at the proteomic and transcriptomic level. Half (19) of these candidates have close functional homologs in other species. Another 12 candidates have homologs designated as hypothetical proteins in other species. The remaining six candidates are short (< 100 AA) and are most likely bona fide novel proteins.

Conclusions

The aggregation of PSM quality information for predicted ORFs provides a robust and efficient method to identify novel proteins in proteomics data. The workflow is in particular capable of identifying small proteins and frameshift variants. Since PSMs are explicitly mapped to genomic locations, it furthermore facilitates the integration of transcriptomics data and other sources of genome-level information.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12859-021-04159-8.

Proteogenomics Study of Blastobotrys adeninivorans TMCC 70007-A Dominant Yeast in the Fermentation Process of Pu-erh Tea

Blastobotrys adeninivorans plays an essential role in pile-fermenting of Pu-erh tea. Its ability to assimilate various carbon and nitrogen sources makes it available for application in a wide range of industry sectors. The genome of B. adeninivorans TMCC 70007 isolated from pile-fermented Pu-erh tea was sequenced and assembled. Proteomics analysis indicated that 4900 proteins in TMCC 70007 were expressed under various culture conditions. Proteogenomics mapping revealed 48 previously unknown genes and corrected 118 gene models predicted by GeneMark-ES. Ortho-proteogenomics analysis identified 17 previously unidentified genes in B. adeninivorans LS3, the first strain with a sequenced genome among the genus Blastobotrys as well. More importantly, five species specific genes were identified from TMCC 70007, which could serve as a barcode for strain typing and were applicable for fermentation process protection of this industrial species. The datasets generated from tea aqueous extract culture not only increased the proteome coverage and accuracy but also contributed to the identification of proteins related to polyphenols and caffeine, which were considered to change greatly during the microbial fermentation of Pu-erh tea. This study provides a proteome perspective on TMCC 70007, which was considered to be an important strain in the production of Pu-erh tea. The systematic proteogenomics analysis not only made a better annotation on the genome of B. adeninivorans TMCC 70007 as previous proteogenomics study but also provided solution for fermentation process protection on valuable industrial species with species specific genes uniquely identified from proteogenomics study.

GenTree, an integrated resource for analyzing the evolution and function of primate-specific coding genes

The origination of new genes contributes to phenotypic evolution in humans. Two major challenges in the study of new genes are the inference of gene ages and annotation of their protein-coding potential. To tackle these challenges, we created GenTree, an integrated online database that compiles age inferences from three major methods together with functional genomic data for new genes. Genome-wide comparison of the age inference methods revealed that the synteny-based pipeline (SBP) is most suited for recently duplicated genes, whereas the protein-family–based methods are useful for ancient genes. For SBP-dated primate-specific protein-coding genes (PSGs), we performed manual evaluation based on published PSG lists and showed that SBP generated a conservative data set of PSGs by masking less reliable syntenic regions. After assessing the coding potential based on evolutionary constraint and peptide evidence from proteomic data, we curated a list of 254 PSGs with different levels of protein evidence. This list also includes 41 candidate misannotated pseudogenes that encode primate-specific short proteins. Coexpression analysis showed that PSGs are preferentially recruited into organs with rapidly evolving pathways such as spermatogenesis, immune response, mother–fetus interaction, and brain development. For brain development, primate-specific KRAB zinc-finger proteins (KZNFs) are specifically up-regulated in the mid-fetal stage, which may have contributed to the evolution of this critical stage. Altogether, hundreds of PSGs are either recruited to processes under strong selection pressure or to processes supporting an evolving novel organ.

Discovery of coding regions in the human genome by integrated proteogenomics analysis workflow

Proteogenomics enable the discovery of novel peptides (from unannotated genomic protein-coding loci) and single amino acid variant peptides (derived from single-nucleotide polymorphisms and mutations). Increasing the reliability of these identifications is crucial to ensure their usefulness for genome annotation and potential application as neoantigens in cancer immunotherapy. We here present integrated proteogenomics analysis workflow (IPAW), which combines peptide discovery, curation, and validation. IPAW includes the SpectrumAI tool for automated inspection of MS/MS spectra, eliminating false identifications of single-residue substitution peptides. We employ IPAW to analyze two proteomics data sets acquired from A431 cells and five normal human tissues using extended (pH range, 3–10) high-resolution isoelectric focusing (HiRIEF) pre-fractionation and TMT-based peptide quantitation. The IPAW results provide evidence for the translation of pseudogenes, lncRNAs, short ORFs, alternative ORFs, N-terminal extensions, and intronic sequences. Moreover, our quantitative analysis indicates that protein production from certain pseudogenes and lncRNAs is tissue specific.

Proteogenomics enables the discovery of protein coding regions and disease-relevant mutations but their verification remains challenging. Here, the authors combine peptide discovery, curation and validation in an integrated proteogenomics workflow, robustly identifying unknown coding regions and mutations.

Proteogenomic Investigation of Strain Variation in Clinical Mycobacterium tuberculosis Isolates

Mycobacterium tuberculosis consists of a large number of different strains that display unique virulence characteristics. Whole-genome sequencing has revealed substantial genetic diversity among clinical M. tuberculosis isolates, and elucidating the phenotypic variation encoded by this genetic diversity will be of the utmost importance to fully understand M. tuberculosis biology and pathogenicity. In this study, we integrated whole-genome sequencing and mass spectrometry (GeLC-MS/MS) to reveal strain-specific characteristics in the proteomes of two clinical M. tuberculosis Latin American-Mediterranean isolates. Using this approach, we identified 59 peptides containing single amino acid variants, which covered ∼9% of all coding nonsynonymous single nucleotide variants detected by whole-genome sequencing. Furthermore, we identified 29 distinct peptides that mapped to a hypothetical protein not present in the M. tuberculosis H37Rv reference proteome. Here, we provide evidence for the expression of this protein in the clinical M. tuberculosis SAWC3651 isolate. The strain-specific databases enabled confirmation of genomic differences (i.e., large genomic regions of difference and nonsynonymous single nucleotide variants) in these two clinical M. tuberculosis isolates and allowed strain differentiation at the proteome level. Our results contribute to the growing field of clinical microbial proteogenomics and can improve our understanding of phenotypic variation in clinical M. tuberculosis isolates.

Systematic Comparison of False-Discovery-Rate-Controlling Strategies for Proteogenomic Search Using Spike-in Experiments

Proteogenomic searches are useful for novel peptide identification from tandem mass spectra. Usually, separate and multistage approaches are adopted to accurately control the false discovery rate (FDR) for proteogenomic search. Their performance on novel peptide identification has not been thoroughly evaluated, however, mainly due to the difficulty in confirming the existence of identified novel peptides. We simulated a proteogenomic search using a controlled, spike-in proteomic data set. After confirming that the results of the simulated proteogenomic search were similar to those of a real proteogenomic search using a human cell line data set, we evaluated the performance of six FDR control methods-global, separate, and multistage FDR estimation, respectively, coupled to a target-decoy search and a mixture model-based method-on novel peptide identification. The multistage approach showed the highest accuracy for FDR estimation. However, global and separate FDR estimation with the mixture model-based method showed higher sensitivities than others at the same true FDR. Furthermore, the mixture model-based method performed equally well when applied without or with a reduced set of decoy sequences. Considering different prior probabilities for novel and known protein identification, we recommend using mixture model-based methods with separate FDR estimation for sensitive and reliable identification of novel peptides from proteogenomic searches.

N-terminal Proteomics Assisted Profiling of the Unexplored Translation Initiation Landscape in Arabidopsis thaliana

Proteogenomics is an emerging research field yet lacking a uniform method of analysis. Proteogenomic studies in which N-terminal proteomics and ribosome profiling are combined, suggest that a high number of protein start sites are currently missing in genome annotations. We constructed a proteogenomic pipeline specific for the analysis of N-terminal proteomics data, with the aim of discovering novel translational start sites outside annotated protein coding regions. In summary, unidentified MS/MS spectra were matched to a specific N-terminal peptide library encompassing protein N termini encoded in the Arabidopsis thaliana genome. After a stringent false discovery rate filtering, 117 protein N termini compliant with N-terminal methionine excision specificity and indicative of translation initiation were found. These include N-terminal protein extensions and translation from transposable elements and pseudogenes. Gene prediction provided supporting protein-coding models for approximately half of the protein N termini. Besides the prediction of functional domains (partially) contained within the newly predicted ORFs, further supporting evidence of translation was found in the recently released Araport11 genome re-annotation of Arabidopsis and computational translations of sequences stored in public repositories. Most interestingly, complementary evidence by ribosome profiling was found for 23 protein N termini. Finally, by analyzing protein N-terminal peptides, an in silico analysis demonstrates the applicability of our N-terminal proteogenomics strategy in revealing protein-coding potential in species with well- and poorly-annotated genomes.

Reannotation of Genomes by Means of Proteomics Data

Omics approaches have become popular in biology as powerful discovery tools, and currently gain in interest for diagnostic applications. Establishing the accurate genome sequence of any organism is easy, but the outcome of its annotation by means of automatic pipelines remains imprecise. Some protein-encoding genes may be missed as soon as they are specific and poorly conserved in a given taxon, while important to explain the specific traits of the organism. Translational starts are also poorly predicted in a relatively important number of cases, thus impacting the protein sequence database used in proteomics, comparative genomics, and systems biology. The use of high-throughput proteomics data to improve genome annotation is an attractive option to obtain a more comprehensive molecular picture of a given organism. Here, protocols for reannotating prokaryote genomes are described based on shotgun proteomics and derivatization of protein N-termini with a positively charged reagent coupled to high-resolution tandem mass spectrometry.

GAPP: A Proteogenomic Software for Genome Annotation and Global Profiling of Post-translational Modifications in Prokaryotes

Although the number of sequenced prokaryotic genomes is growing rapidly, experimentally verified annotation of prokaryotic genome remains patchy and challenging. To facilitate genome annotation efforts for prokaryotes, we developed an open source software called GAPP for genome annotation and global profiling of post-translational modifications (PTMs) in prokaryotes. With a single command, it provides a standard workflow to validate and refine predicted genetic models and discover diverse PTM events. We demonstrated the utility of GAPP using proteomic data from Helicobacter pylori, one of the major human pathogens that is responsible for many gastric diseases. Our results confirmed 84.9% of the existing predicted H. pylori proteins, identified 20 novel protein coding genes, and corrected four existing gene models with regard to translation initiation sites. In particular, GAPP revealed a large repertoire of PTMs using the same proteomic data and provided a rich resource that can be used to examine the functions of reversible modifications in this human pathogen. This software is a powerful tool for genome annotation and global discovery of PTMs and is applicable to any sequenced prokaryotic organism; we expect that it will become an integral part of ongoing genome annotation efforts for prokaryotes. GAPP is freely available at https://sourceforge.net/projects/gappproteogenomic/.

Improving GENCODE reference gene annotation using a high-stringency proteogenomics workflow

Complete annotation of the human genome is indispensable for medical research. The GENCODE consortium strives to provide this, augmenting computational and experimental evidence with manual annotation. The rapidly developing field of proteogenomics provides evidence for the translation of genes into proteins and can be used to discover and refine gene models. However, for both the proteomics and annotation groups, there is a lack of guidelines for integrating this data. Here we report a stringent workflow for the interpretation of proteogenomic data that could be used by the annotation community to interpret novel proteogenomic evidence. Based on reprocessing of three large-scale publicly available human data sets, we show that a conservative approach, using stringent filtering is required to generate valid identifications. Evidence has been found supporting 16 novel protein-coding genes being added to GENCODE. Despite this many peptide identifications in pseudogenes cannot be annotated due to the absence of orthogonal supporting evidence.

Identifying and annotating functional elements in the human genome remains a challenging but important task. Here the authors propose a priority annotation score to rank identifications and suggest how proteogenomics evidence can be interpreted and what additional information substantiates protein-coding potential for annotation.

The potential clinical impact of the release of two drafts of the human proteome

The authors have carried out an investigation of the two "draft maps of the human proteome" published in 2014 in Nature. The findings include an abundance of poor spectra, low-scoring peptide-spectrum matches and incorrectly identified proteins in both these studies, highlighting clear issues with the application of false discovery rates. This noise means that the claims made by the two papers - the identification of high numbers of protein coding genes, the detection of novel coding regions and the draft tissue maps themselves - should be treated with considerable caution. The authors recommend that clinicians and researchers do not use the unfiltered data from these studies. Despite this these studies will inspire further investigation into tissue-based proteomics. As long as this future work has proper quality controls, it could help produce a consensus map of the human proteome and improve our understanding of the processes that underlie health and disease.