Computational prediction of proteotypic peptides

Evaluation of: Mallick P, Schirle M, Chen SS et al. Computational prediction of proteotypic peptides for quantitative proteomics. Nat. Biotechnol. 25(1), 125–131 (2007).

Mass spectrometry, the driving analytical force behind proteomics, is primarily used to identify and quantify as many proteins in a complex biological mixture as possible. While there are many ways to prepare samples, one aspect that is common to a vast majority of bottom-up proteomic studies is the digestion of proteins into tryptic peptides prior to their analysis by mass spectrometry. As correctly highlighted by Mallick and colleagues, only a few peptides are repeatedly and consistently identified for any given protein within a complex mixture. While the existence of these proteotypic peptides (to borrow the authors’ terminology) is well known in the proteomics community, there has never been an empirical method to recognize which peptides may be proteotypic for a given protein. In this study, the investigators discovered over 16,000 proteotypic peptides from a collection of over 600,000 peptide identifications obtained from four different analytical platforms. The study examined a number of physicochemical parameters of these peptides to determine which properties were most relevant in defining a proteotypic peptide. These characteristic properties were then used to develop computational tools to predict proteotypic peptides for any given protein within an organism.     

Computational prediction of proteotypic peptides for quantitative proteomics

Mass spectrometry-based quantitative proteomics has become an important component of biological and clinical research. Although such analyses typically assume that a protein's peptide fragments are observed with equal likelihood, only a few so-called 'proteotypic' peptides are repeatedly and consistently identified for any given protein present in a mixture. Using >600,000 peptide identifications generated by four proteomic platforms, we empirically identified >16,000 proteotypic peptides for 4,030 distinct yeast proteins. Characteristic physicochemical properties of these peptides were used to develop a computational tool that can predict proteotypic peptides for any protein from any organism, for a given platform, with >85% cumulative accuracy. Possible applications of proteotypic peptides include validation of protein identifications, absolute quantification of proteins, annotation of coding sequences in genomes, and characterization of the physical principles governing key elements of mass spectrometric workflows (e.g., digestion, chromatography, ionization and fragmentation).     

High quality catalog of proteotypic peptides from human heart

Proteomics research is beginning to expand beyond the more traditional shotgun analysis of protein mixtures to include targeted analyses of specific proteins using mass spectrometry. Integral to the development of a robust assay based on targeted mass spectrometry is prior knowledge of which peptides provide an accurate and sensitive proxy of the originating gene product (i.e., proteotypic peptides). To develop a catalog of "proteotypic peptides" in human heart, TRIzol extracts of left-ventricular tissue from nonfailing and failing human heart explants were optimized for shotgun proteomic analysis using Multidimensional Protein Identification Technology (MudPIT). Ten replicate MudPIT analyses were performed on each tissue sample and resulted in the identification of 30 605 unique peptides with a q-value < or = 0.01, corresponding to 7138 unique human heart proteins. Experimental observation frequencies were assessed and used to select over 4476 proteotypic peptides for 2558 heart proteins. This human cardiac data set can serve as a public reference to guide the selection of proteotypic peptides for future targeted mass spectrometry experiments monitoring potential protein biomarkers of human heart diseases.     

Assigning statistical significance to proteotypic peptides via database searches

Querying MS/MS spectra against a database containing only proteotypic peptides reduces data analysis time due to reduction of database size. Despite the speed advantage, this search strategy is challenged by issues of statistical significance and coverage. The former requires separating systematically significant identifications from less confident identifications, while the latter arises when the underlying peptide is not present, due to single amino acid polymorphisms (SAPs) or post-translational modifications (PTMs), in the proteotypic peptide libraries searched. To address both issues simultaneously, we have extended RAId's knowledge database to include proteotypic information, utilized RAId's statistical strategy to assign statistical significance to proteotypic peptides, and modified RAId's programs to allow for consideration of proteotypic information during database searches. The extended database alleviates the coverage problem since all annotated modifications, even those that occurred within proteotypic peptides, may be considered. Taking into account the likelihoods of observation, the statistical strategy of RAId provides accurate E-value assignments regardless whether a candidate peptide is proteotypic or not. The advantage of including proteotypic information is evidenced by its superior retrieval performance when compared to regular database searches.     

Computational binding assays of antigenic peptides

Summary Computer models can be combined with laboratory experiments for the efficient determination of (i) peptides that bind MHC molecules and (ii) T-cell epitopes. For maximum benefit, the use of computer models must be treated as experiments analogous to standard laboratory procedures. This requires the definition of standards and experimental protocols for model application. We describe the requirements for validation and assessment of computer models. The utility of combining accurate predictions with a limited number of laboratory experiments is illustrated by practical examples. These include the identification of T-cell epitopes from IDDM-, melanoma-and malaria-related antigens by combining computational and conventional laboratory assays. The success rate in determining antigenic peptides, each in the context of a specific HLA molecule, ranged from 27 to 71%, while the natural prevalence of MHC-binding peptides is 0.1–5%.     

Computational binding assays of antigenic peptides

Computer models can be combined with laboratory experiments for the efficient determination of (i) peptides that bind MHC molecules and (ii) T-cell epitopes. For maximum benefit, the use of computer models must be treated as experiments analogous to standard laboratory procedures. This requires the definition of standards and experimental protocols for model application. We describe the requirements for validation and assessment of computer models. The utility of combining accurate predictions with a limited number of laboratory experiments is illustrated by practical examples. These include the identification of T-cell epitopes from IDDM-, melanoma- and malaria-related antigens by combining computational and conventional laboratory assays. The success rate in determining antigenic peptides, each in the context of a specific HLA molecule, ranged from 27 to 71%, while the natural prevalence of MHC-binding peptides is 0.1–5%.     

Computational prediction of protein-protein complexes

ABSTRACT: BACKGROUND: Protein-protein interactions form the core of several biological processes. With protein-protein interfaces being considered as drug targets, studies on their interactions and molecular mechanisms are gaining ground. As the number of protein complexes in databases is scarce as compared to a spectrum of independent protein molecules, computational approaches are being considered for speedier model derivation and assessment of a plausible complex. In this study, a good approach towards in silico generation of protein-protein heterocomplex and identification of the most probable complex among thousands of complexes thus generated is documented. This approach becomes even more useful in the event of little or no binding site information between the interacting protein molecules. FINDINGS: A plausible protein-protein hetero-complex was fished out from 10 docked complexes which are a representative set of complexes obtained after clustering of 2000 generated complexes using protein-protein docking softwares. The interfacial area for this complex was predicted by two "hotspot" prediction programs employing different algorithms. Further, this complex had the lowest energy and most buried surface area of all the complexes with the same interfacial residues. CONCLUSIONS: For the generation of a plausible protein heterocomplex, various software tools were employed. Prominent are the protein-protein docking methods, prediction of 'hotspots' which are the amino acid residues likely to be in an interface and measurement of buried surface area of the complexes. Consensus generated in their predictions lends credence to the use of the various softwares used.     

Computational approaches to gene prediction

The problems associated with gene identification and the prediction of gene structure in DNA sequences have been the focus of increased attention over the past few years with the recent acquisition by large-scale sequencing projects of an immense amount of genome data. A variety of prediction programs have been developed in order to address these problems. This paper presents a review of the computational approaches and gene-finders used commonly for gene prediction in eukaryotic genomes. Two approaches, in general, have been adopted for this purpose: similarity-based and ab initio techniques. The information gleaned from these methods is then combined via a variety of algorithms, including Dynamic Programming (DP) or the Hidden Markov Model (HMM), and then used for gene prediction from the genomic sequences.     

PEPOP: Computational design of immunogenic peptides

Background  

Most methods available to predict protein epitopes are sequence based. There is a need for methods using 3D information for prediction of discontinuous epitopes and derived immunogenic peptides.     

Computational prediction of eukaryotic protein-coding genes

The human genome sequence is the book of our life. Buried in this large volume are our genes, which are scattered as small DNA fragments throughout the genome and comprise a small percentage of the total text. Finding these indistinct 'needles' in a vast genomic 'haystack' can be extremely challenging. In response to this challenge, computational prediction approaches have proliferated in recent years that predict the location and structure of genes. Here, I discuss these approaches and explain why they have become essential for the analyses of newly sequenced genomes.     

Computational prediction of host-pathogen protein-protein interactions

MOTIVATION: Infectious diseases such as malaria result in millions of deaths each year. An important aspect of any host-pathogen system is the mechanism by which a pathogen can infect its host. One method of infection is via protein-protein interactions (PPIs) where pathogen proteins target host proteins. Developing computational methods that identify which PPIs enable a pathogen to infect a host has great implications in identifying potential targets for therapeutics. RESULTS: We present a method that integrates known intra-species PPIs with protein-domain profiles to predict PPIs between host and pathogen proteins. Given a set of intra-species PPIs, we identify the functional domains in each of the interacting proteins. For every pair of functional domains, we use Bayesian statistics to assess the probability that two proteins with that pair of domains will interact. We apply our method to the Homo sapiens-Plasmodium falciparum host-pathogen system. Our system predicts 516 PPIs between proteins from these two organisms. We show that pairs of human proteins we predict to interact with the same Plasmodium protein are close to each other in the human PPI network and that Plasmodium pairs predicted to interact with same human protein are co-expressed in DNA microarray datasets measured during various stages of the Plasmodium life cycle. Finally, we identify functionally enriched sub-networks spanned by the predicted interactions and discuss the plausibility of our predictions. AVAILABILITY: Supplementary data are available at http://staff.vbi.vt.edu/dyermd/publications/dyer2007a.html. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.     

Computational prediction of RNA editing sites

MOTIVATION: Some organisms edit their messenger RNA resulting in differences between the genomic sequence for a gene and the corresponding messenger RNA sequence. This difference complicates experimental and computational attempts to find and study genes in organisms with RNA editing even if the full genomic sequence is known. Nevertheless, knowledge of these editing sites is crucial for understanding the editing machinery of these organisms. RESULTS: We present a computational technique that predicts the position of editing sites in the genomic sequence. It uses a statistical approach drawing on the protein sequences of related genes and general features of editing sites of the organism. We apply the method to the mitochondrion of the slime mold Physarum polycephalum. It correctly predicts over 90% of the amino acids and over 70% of the editing sites.     

Computational resources and tools for antimicrobial peptides

Antimicrobial peptides (AMPs), as evolutionarily conserved components of innate immune system, protect against pathogens including bacteria, fungi, viruses, and parasites. In general, AMPs are relatively small peptides (<10 kDa) with cationic nature and amphipathic structure and have modes of action different from traditional antibiotics. Up to now, there are more than 19 000 AMPs that have been reported, including those isolated from nature sources or by synthesis. They have been considered to be promising substitutes of conventional antibiotics in the quest to address the increasing occurrence of antibiotic resistance. However, most AMPs have modest direct antimicrobial activity, and their mechanisms of action, as well as their structure–activity relationships, are still poorly understood. Computational strategies are invaluable assets to provide insight into the activity of AMPs and thus exploit their potential as a new generation of antimicrobials. This article reviews the advances of AMP databases and computational tools for the prediction and design of new active AMPs. Copyright © 2016 European Peptide Society and John Wiley & Sons, Ltd.     

Computational identification of self-inhibitory peptides from envelope proteins

Fusion process is known to be the initial step of viral infection and hence targeting the entry process is a promising strategy to design antiviral therapy. The self-inhibitory peptides derived from the enveloped (E) proteins function to inhibit the protein-protein interactions in the membrane fusion step mediated by the viral E protein. Thus, they have the potential to be developed into effective antiviral therapy. Herein, we have developed a Monte Carlo-based computational method with the aim to identify and optimize potential peptide hits from the E proteins. The stability of the peptides, which indicates their potential to bind in situ to the E proteins, was evaluated by two different scoring functions, dipolar distance-scaled, finite, ideal-gas reference state and residue-specific all-atom probability discriminatory function. The method was applied to α-helical Class I HIV-1 gp41, β-sheet Class II Dengue virus (DENV) type 2 E proteins, as well as Class III Herpes Simplex virus-1 (HSV-1) glycoprotein, a E protein with a mixture of α-helix and β-sheet structural fold. The peptide hits identified are in line with the druggable regions where the self-inhibitory peptide inhibitors for the three classes of viral fusion proteins were derived. Several novel peptides were identified from either the hydrophobic regions or the functionally important regions on Class II DENV-2 E protein and Class III HSV-1 gB. They have potential to disrupt the protein-protein interaction in the fusion process and may serve as starting points for the development of novel inhibitors for viral E proteins.     

Computational methods in protein structure prediction

This review presents the advances in protein structure prediction from the computational methods perspective. The approaches are classified into four major categories: comparative modeling, fold recognition, first principles methods that employ database information, and first principles methods without database information. Important advances along with current limitations and challenges are presented.