Towards low false discovery rate estimation for protein-protein interactions detected by chemical cross-linking.

Chemical cross-linking (CX) of proteins in vivo or in cell free extracts followed by mass spectrometric (MS) identification of linked peptide pairs (CXMS) can reveal protein-protein interactions (PPIs) both at a proteome wide scale and the level of cross-linked amino acid residues. However, error estimation at the level of PPI remains challenging in large scale datasets. Here we discuss recent advances in the recognition of spurious inter-protein peptide pairs and in diminishing the FDR for these PPI-signaling cross-links, such as the use of chromatographic retention time prediction, in order to come to a more reliable reporting of PPIs.

Misuse of the Michaelis-Menten rate law for protein interaction networks and its remedy.

For over a century, the Michaelis-Menten (MM) rate law has been used to describe the rates of enzyme-catalyzed reactions and gene expression. Despite the ubiquity of the MM rate law, it accurately captures the dynamics of underlying biochemical reactions only so long as it is applied under the right condition, namely, that the substrate is in large excess over the enzyme-substrate complex. Unfortunately, in circumstances where its validity condition is not satisfied, especially so in protein interaction networks, the MM rate law has frequently been misused. In this review, we illustrate how inappropriate use of the MM rate law distorts the dynamics of the system, provides mistaken estimates of parameter values, and makes false predictions of dynamical features such as ultrasensitivity, bistability, and oscillations. We describe how these problems can be resolved with a slightly modified form of the MM rate law, based on the total quasi-steady state approximation (tQSSA). Furthermore, we show that the tQSSA can be used for accurate stochastic simulations at a lower computational cost than using the full set of mass-action rate laws. This review describes how to use quasi-steady state approximations in the right context, to prevent drawing erroneous conclusions from in silico simulations.

Prediction of Protein-Protein Interaction Sites Using Convolutional Neural Network and Improved Data Sets.

Protein-protein interaction (PPI) sites play a key role in the formation of protein complexes, which is the basis of a variety of biological processes. Experimental methods to solve PPI sites are expensive and time-consuming, which has led to the development of different kinds of prediction algorithms. We propose a convolutional neural network for PPI site prediction and use residue binding propensity to improve the positive samples. Our method obtains a remarkable result of the area under the curve (AUC) = 0.912 on the improved data set. In addition, it yields much better results on samples with high binding propensity than on randomly selected samples. This suggests that there are considerable false-positive PPI sites in the positive samples defined by the distance between residue atoms.

APEX2-mediated proximity labeling resolves protein networks in Saccharomyces cerevisiae cells.

Enzyme-catalyzed proximity labeling (PL) with the engineered ascorbate peroxidase APEX2 is a novel approach to map organelle compartmentalization and protein networks in living cells. Current procedures developed for mammalian cells do not allow delivery of the cosubstrate, biotin-phenol, into living yeast cells. Here, we present a new method based on semipermeabilized yeast cells. Combined with stable isotope labeling by amino acids in cell culture (SILAC), we demonstrate proteomic mapping of a membrane-enclosed and a semiopen compartment, the mitochondrial matrix and the nucleus. APEX2 PL revealed nuclear proteins that were previously not identified by conventional techniques. One of these, the Yer156C protein, is highly conserved but of unknown function. Its human ortholog, melanocyte proliferating gene 1, is linked to developmental processes and dermatological diseases. A first characterization of the Yer156C neighborhood reveals an array of proteins linked to proteostasis and RNA binding. Thus, our approach establishes APEX2 PL as another powerful tool that complements the methods palette for the model system yeast.

Identification of Essential Proteins Based on Improved HITS Algorithm.

Essential proteins are critical to the development and survival of cells. Identifying and analyzing essential proteins is vital to understand the molecular mechanisms of living cells and design new drugs. With the development of high-throughput technologies, many protein⁻protein interaction (PPI) data are available, which facilitates the studies of essential proteins at the network level. Up to now, although various computational methods have been proposed, the prediction precision still needs to be improved. In this paper, we propose a novel method by applying Hyperlink-Induced Topic Search (HITS) on weighted PPI networks to detect essential proteins, named HSEP. First, an original undirected PPI network is transformed into a bidirectional PPI network. Then, both biological information and network topological characteristics are taken into account to weighted PPI networks. Pieces of biological information include gene expression data, Gene Ontology (GO) annotation and subcellular localization. The edge clustering coefficient is represented as network topological characteristics to measure the closeness of two connected nodes. We conducted experiments on two species, namely Saccharomyces cerevisiae and Drosophila melanogaster, and the experimental results show that HSEP outperformed some state-of-the-art essential proteins detection techniques.

A Novel Method for Identifying Essential Genes by Fusing Dynamic Protein⁻Protein Interactive Networks.

Essential genes play an indispensable role in supporting the life of an organism. Identification of essential genes helps us to understand the underlying mechanism of cell life. The essential genes of bacteria are potential drug targets of some diseases genes. Recently, several computational methods have been proposed to detect essential genes based on the static protein⁻protein interactive (PPI) networks. However, these methods have ignored the fact that essential genes play essential roles under certain conditions. In this work, a novel method was proposed for the identification of essential proteins by fusing the dynamic PPI networks of different time points (called by FDP). Firstly, the active PPI networks of each time point were constructed and then they were fused into a final network according to the networks' similarities. Finally, a novel centrality method was designed to assign each gene in the final network a ranking score, whilst considering its orthologous property and its global and local topological properties in the network. This model was applied on two different yeast data sets. The results showed that the FDP achieved a better performance in essential gene prediction as compared to other existing methods that are based on the static PPI network or that are based on dynamic networks.

IID 2018 update: context-specific physical protein-protein interactions in human, model organisms and domesticated species.

Knowing the set of physical protein-protein interactions (PPIs) that occur in a particular context-a tissue, disease, or other condition-can provide valuable insights into key research questions. However, while the number of identified human PPIs is expanding rapidly, context information remains limited, and for most non-human species context-specific networks are completely unavailable. The Integrated Interactions Database (IID) provides one of the most comprehensive sets of context-specific human PPI networks, including networks for 133 tissues, 91 disease conditions, and many other contexts. Importantly, it also provides context-specific networks for 17 non-human species including model organisms and domesticated animals. These species are vitally important for drug discovery and agriculture. IID integrates interactions from multiple databases and datasets. It comprises over 4.8 million PPIs annotated with several types of context: tissues, subcellular localizations, diseases, and druggability information (the latter three are new annotations not available in the previous version). This update increases the number of species from 6 to 18, the number of PPIs from ∼1.5 million to ∼4.8 million, and the number of tissues from 30 to 133. IID also now supports topology and enrichment analyses of returned networks. IID is available at http://ophid.utoronto.ca/iid.

The Impact of Protein Structure and Sequence Similarity on the Accuracy of Machine-Learning Scoring Functions for Binding Affinity Prediction.

It has recently been claimed that the outstanding performance of machine-learning scoring functions (SFs) is exclusively due to the presence of training complexes with highly similar proteins to those in the test set. Here, we revisit this question using 24 similarity-based training sets, a widely used test set, and four SFs. Three of these SFs employ machine learning instead of the classical linear regression approach of the fourth SF (X-Score which has the best test set performance out of 16 classical SFs). We have found that random forest (RF)-based RF-Score-v3 outperforms X-Score even when 68% of the most similar proteins are removed from the training set. In addition, unlike X-Score, RF-Score-v3 is able to keep learning with an increasing training set size, becoming substantially more predictive than X-Score when the full 1105 complexes are used for training. These results show that machine-learning SFs owe a substantial part of their performance to training on complexes with dissimilar proteins to those in the test set, against what has been previously concluded using the same data. Given that a growing amount of structural and interaction data will be available from academic and industrial sources, this performance gap between machine-learning SFs and classical SFs is expected to enlarge in the future.

RocSampler: regularizing overlapping protein complexes in protein-protein interaction networks.

BACKGROUND: In recent years, protein-protein interaction (PPI) networks have been well recognized as important resources to elucidate various biological processes and cellular mechanisms. In this paper, we address the problem of predicting protein complexes from a PPI network. This problem has two difficulties. One is related to small complexes, which contains two or three components. It is relatively difficult to identify them due to their simpler internal structure, but unfortunately complexes of such sizes are dominant in major protein complex databases, such as CYC2008. Another difficulty is how to model overlaps between predicted complexes, that is, how to evaluate different predicted complexes sharing common proteins because CYC2008 and other databases include such protein complexes. Thus, it is critical how to model overlaps between predicted complexes to identify them simultaneously. RESULTS: In this paper, we propose a sampling-based protein complex prediction method, RocSampler (Regularizing Overlapping Complexes), which exploits, as part of the whole scoring function, a regularization term for the overlaps of predicted complexes and that for the distribution of sizes of predicted complexes. We have implemented RocSampler in MATLAB and its executable file for Windows is available at the site, http://imi.kyushu-u.ac.jp/~om/software/RocSampler/ . CONCLUSIONS: We have applied RocSampler to five yeast PPI networks and shown that it is superior to other existing methods. This implies that the design of scoring functions including regularization terms is an effective approach for protein complex prediction.

Detection of Complexes in Biological Networks Through Diversified Dense Subgraph Mining.

Protein-protein interaction (PPI) networks, providing a comprehensive landscape of protein interaction patterns, enable us to explore biological processes and cellular components at multiple resolutions. For a biological process, a number of proteins need to work together to perform a job. Proteins densely interact with each other, forming large molecular machines or cellular building blocks. Identification of such densely interconnected clusters or protein complexes from PPI networks enables us to obtain a better understanding of the hierarchy and organization of biological processes and cellular components. However, most existing graph clustering algorithms on PPI networks often cannot effectively detect densely connected subgraphs and overlapped subgraphs. In this article, we formulate the problem of complex detection as diversified dense subgraph mining and introduce a novel approximation algorithm to efficiently enumerate putative protein complexes from biological networks. The key insight of our algorithm is that instead of enumerating all dense subgraphs, we only need to find a small diverse subset of subgraphs that cover as many proteins as possible. The problem is modeled as finding a diverse set of maximal dense subgraphs where we develop highly effective pruning techniques to guarantee efficiency. To scale up to large networks, we devise a divide-and-conquer approach to speed up the algorithm in a distributed manner. By comparing with existing clustering and dense subgraph-based algorithms on several yeast and human PPI networks, we demonstrate that our method can detect more putative protein complexes and achieves better prediction accuracy.

SEQUOIA: significance enhanced network querying through context-sensitive random walk and minimization of network conductance.

BACKGROUND: Network querying algorithms provide computational means to identify conserved network modules in large-scale biological networks that are similar to known functional modules, such as pathways or molecular complexes. Two main challenges for network querying algorithms are the high computational complexity of detecting potential isomorphism between the query and the target graphs and ensuring the biological significance of the query results. RESULTS: In this paper, we propose SEQUOIA, a novel network querying algorithm that effectively addresses these issues by utilizing a context-sensitive random walk (CSRW) model for network comparison and minimizing the network conductance of potential matches in the target network. The CSRW model, inspired by the pair hidden Markov model (pair-HMM) that has been widely used for sequence comparison and alignment, can accurately assess the node-to-node correspondence between different graphs by accounting for node insertions and deletions. The proposed algorithm identifies high-scoring network regions based on the CSRW scores, which are subsequently extended by maximally reducing the network conductance of the identified subnetworks. CONCLUSIONS: Performance assessment based on real PPI networks and known molecular complexes show that SEQUOIA outperforms existing methods and clearly enhances the biological significance of the query results. The source code and datasets can be downloaded from http://www.ece.tamu.edu/~bjyoon/SEQUOIA .

Reference Tissue-Based Kinetic Evaluation of 18F-AV-1451 for Tau Imaging.

The goal of this paper was to evaluate the in vivo kinetics of the novel tau-specific PET radioligand 18F-AV-1451 in cognitively healthy control (HC) and Alzheimer disease (AD) subjects, using reference region analyses. METHODS: 18F-AV-1451 PET imaging was performed on 43 subjects (5 young HCs, 23 older HCs, and 15 AD subjects). Data were collected from 0 to 150 min after injection, with a break from 100 to 120 min. T1-weighted MR images were segmented using FreeSurfer to create 14 bilateral regions of interest (ROIs). In all analyses, cerebellar gray matter was used as the reference region. Nondisplaceable binding potentials (BPNDs) were calculated using the simplified reference tissue model (SRTM) and SRTM2; the Logan graphical analysis distribution volume ratio (DVR) was calculated for 30-150 min (DVR30-150). These measurements were compared with each other and used as reference standards for defining an appropriate 20-min window for the SUV ratio (SUVR). Pearson correlations were used to compare the reference standards to 20-min SUVRs (start times varied from 30 to 130 min), for all values, for ROIs with low 18F-AV-1451 binding (lROIs, mean of BPND + 1 and DVR30-150 < 1.5), and for ROIs with high 18F-AV-1451 binding (hROIs, mean of BPND + 1 and DVR30-150 > 1.5). RESULTS: SRTM2 BPND + 1 and DVR30-150 were in good agreement. Both were in agreement with SRTM BPND + 1 for lROIs but were greater than SRTM BPND + 1 for hROIs, resulting in a nonlinear relationship. hROI SUVRs increased from 80-100 to 120-140 min by 0.24 ± 0.15. The SUVR time interval resulting in the highest correlation and slope closest to 1 relative to the reference standards for all values was 120-140 min for hROIs, 60-80 min for lROIs, and 80-100 min for lROIs and hROIs. There was minimal difference between methods when statistical significance between ADs and HCs was calculated. CONCLUSION: Despite later time periods providing better agreement between reference standards and SUVRs for hROIs, a good compromise for studying lROIs and hROIs is SUVR80-100. The lack of SUVR plateau for hROIs highlights the importance of precise acquisition time for longitudinal assessment.

Protein complex analysis: From raw protein lists to protein interaction networks.

The elucidation of molecular interaction networks is one of the pivotal challenges in the study of biology. Affinity purification-mass spectrometry and other co-complex methods have become widely employed experimental techniques to identify protein complexes. These techniques typically suffer from a high number of false negatives and false positive contaminants due to technical shortcomings and purification biases. To support a diverse range of experimental designs and approaches, a large number of computational methods have been proposed to filter, infer and validate protein interaction networks from experimental pull-down MS data. Nevertheless, this expansion of available methods complicates the selection of the most optimal ones to support systems biology-driven knowledge extraction. In this review, we give an overview of the most commonly used computational methods to process and interpret co-complex results, and we discuss the issues and unsolved problems that still exist within the field. © 2015 Wiley Periodicals, Inc. Mass Spec Rev 36:600-614, 2017.

APID interactomes: providing proteome-based interactomes with controlled quality for multiple species and derived networks.

APID (Agile Protein Interactomes DataServer) is an interactive web server that provides unified generation and delivery of protein interactomes mapped to their respective proteomes. This resource is a new, fully redesigned server that includes a comprehensive collection of protein interactomes for more than 400 organisms (25 of which include more than 500 interactions) produced by the integration of only experimentally validated protein-protein physical interactions. For each protein-protein interaction (PPI) the server includes currently reported information about its experimental validation to allow selection and filtering at different quality levels. As a whole, it provides easy access to the interactomes from specific species and includes a global uniform compendium of 90,379 distinct proteins and 678,441 singular interactions. APID integrates and unifies PPIs from major primary databases of molecular interactions, from other specific repositories and also from experimentally resolved 3D structures of protein complexes where more than two proteins were identified. For this purpose, a collection of 8,388 structures were analyzed to identify specific PPIs. APID also includes a new graph tool (based on Cytoscape.js) for visualization and interactive analyses of PPI networks. The server does not require registration and it is freely available for use at http://apid.dep.usal.es.

Standardized Expression of 18F-NAV4694 and 11C-PiB ß-Amyloid PET Results with the Centiloid Scale.

UNLABELLED: A common quantitative output value for PET measures of ß-amyloid (Aß) binding across tracers and methods would allow better comparison of data across sites and application of universal diagnostic and prognostic values. A method has recently been developed that generates a unit of measurement called the centiloid. We applied this method to 2-[2-(18)F-fluoro-6-(methylamino)-3-pyridinyl]-1-benzofuran-5-ol ((18)F-NAV4694) and (11)C-Pittsburgh compound B ((11)C-PiB) Aß images to derive the scaling factor required to express tracer binding in centiloids. METHODS: Fifty-five participants, including 10 young controls (33 ± 7 y old), underwent both (11)C-PiB and (18)F-NAV4694 imaging no more than 3 mo apart, with the images acquired 50-70 min after tracer injection. The images were spatially normalized and analyzed using the standard centiloid method and regions (cortex and whole-cerebellum reference) downloaded from the Global Alzheimer Association Interactive Network website. RESULTS: SUV ratios (SUVRs) showed a strong correlation in tracer binding ((18)F-NAV4694 SUVR = 1.09 × (11)C-PiB SUVR - 0.08, R(2) = 0.99). The equation to convert (18)F-NAV4694 to centiloids [100 × ((18)F-NAV4694 SUVR - 1.028)/1.174] was similar to a published equation for (11)C-PiB [100 × ((11)C-PiB SUVR - 1.009)/1.067]. In the young controls, the variance ratio ((18)F-NAV4694 centiloid SD divided by (11)C-PiB centiloid SD) was 0.85. CONCLUSION: The results for both (11)C-PiB and (18)F-NAV4694 can now be expressed in centiloids, an important step that should allow better clinical and research use of Aß imaging. The standard centiloid method also showed that (18)F-NAV4694 has slightly higher Aß binding and lower variance than (11)C-PiB, important properties for detecting early Aß deposition and change over time.

Development of a Model Protein Interaction Pair as a Benchmarking Tool for the Quantitative Analysis of 2-Site Protein-Protein Interactions.

A significant challenge in the molecular interaction field is to accurately determine the stoichiometry and stepwise binding affinity constants for macromolecules having >1 binding site. The mission of the Molecular Interactions Research Group (MIRG) of the Association of Biomolecular Resource Facilities (ABRF) is to show how biophysical technologies are used to quantitatively characterize molecular interactions, and to educate the ABRF members and scientific community on the utility and limitations of core technologies [such as biosensor, microcalorimetry, or analytic ultracentrifugation (AUC)]. In the present work, the MIRG has developed a robust model protein interaction pair consisting of a bivalent variant of the Bacillus amyloliquefaciens extracellular RNase barnase and a variant of its natural monovalent intracellular inhibitor protein barstar. It is demonstrated that this system can serve as a benchmarking tool for the quantitative analysis of 2-site protein-protein interactions. The protein interaction pair enables determination of precise binding constants for the barstar protein binding to 2 distinct sites on the bivalent barnase binding partner (termed binase), where the 2 binding sites were engineered to possess affinities that differed by 2 orders of magnitude. Multiple MIRG laboratories characterized the interaction using isothermal titration calorimetry (ITC), AUC, and surface plasmon resonance (SPR) methods to evaluate the feasibility of the system as a benchmarking model. Although general agreement was seen for the binding constants measured using solution-based ITC and AUC approaches, weaker affinity was seen for surface-based method SPR, with protein immobilization likely affecting affinity. An analysis of the results from multiple MIRG laboratories suggests that the bivalent barnase-barstar system is a suitable model for benchmarking new approaches for the quantitative characterization of complex biomolecular interactions.

Lensfree optofluidic plasmonic sensor for real-time and label-free monitoring of molecular binding events over a wide field-of-view.

We demonstrate a high-throughput biosensing device that utilizes microfluidics based plasmonic microarrays incorporated with dual-color on-chip imaging toward real-time and label-free monitoring of biomolecular interactions over a wide field-of-view of >20 mm(2). Weighing 40 grams with 8.8 cm in height, this biosensor utilizes an opto-electronic imager chip to record the diffraction patterns of plasmonic nanoapertures embedded within microfluidic channels, enabling real-time analyte exchange. This plasmonic chip is simultaneously illuminated by two different light-emitting-diodes that are spectrally located at the right and left sides of the plasmonic resonance mode, yielding two different diffraction patterns for each nanoaperture array. Refractive index changes of the medium surrounding the near-field of the nanostructures, e.g., due to molecular binding events, induce a frequency shift in the plasmonic modes of the nanoaperture array, causing a signal enhancement in one of the diffraction patterns while suppressing the other. Based on ratiometric analysis of these diffraction images acquired at the detector-array, we demonstrate the proof-of-concept of this biosensor by monitoring in real-time biomolecular interactions of protein A/G with immunoglobulin G (IgG) antibody. For high-throughput on-chip fabrication of these biosensors, we also introduce a deep ultra-violet lithography technique to simultaneously pattern thousands of plasmonic arrays in a cost-effective manner.

Measuring protein interactions using Förster resonance energy transfer and fluorescence lifetime imaging microscopy.

The method of fluorescence lifetime imaging microscopy (FLIM) is a quantitative approach that can be used to detect Förster resonance energy transfer (FRET). The use of FLIM to measure the FRET that results from the interactions between proteins labeled with fluorescent proteins (FPs) inside living cells provides a non-invasive method for mapping interactomes. Here, the use of the phasor plot method to analyze frequency domain (FD) FLIM measurements is described, and measurements obtained from cells producing the 'FRET standard' fusion proteins are used to validate the FLIM system for FRET measurements. The FLIM FRET approach is then used to measure both homologous and heterologous protein-protein interactions (PPI) involving the CCAAT/enhancer-binding protein alpha (C/EBPα). C/EBPα is a transcription factor that controls cell differentiation, and localizes to heterochromatin where it interacts with the heterochromatin protein 1 alpha (HP1α). The FLIM-FRET method is used to quantify the homologous interactions between the FP-labeled basic leucine zipper (BZip) domain of C/EBPα. Then the heterologous interactions between the C/EBPa BZip domain and HP1a are quantified using the FRET-FLIM method. The results demonstrate that the basic region and leucine zipper (BZip) domain of C/EBPα is sufficient for the interaction with HP1α in regions of heterochromatin.

Bioinformatics analysis of protein interaction networks: statistics, topologies, and meeting the standards of experimental biologists.

In recent years, bioinformatics analyses of protein networks have allowed researchers to obtain exceptional theoretical predictions and subsequent experimental confirmations. The current view is that protein networks are scale-free networks and have a topology analogous to that of transport networks, the Internet, and social networks. However, an alternative hypothesis exists in which protein networks and scale-free networks possess significantly different properties. In this work, we show that existing information is insufficient to describe protein networks as scale-free networks.

Improved method for protein complex detection using bottleneck proteins.

BACKGROUND: Detecting protein complexes is one of essential and fundamental tasks in understanding various biological functions or processes. Therefore accurate identification of protein complexes is indispensable. METHODS: For more accurate detection of protein complexes, we propose an algorithm which detects dense protein sub-networks of which proteins share closely located bottleneck proteins. The proposed algorithm is capable of finding protein complexes which allow overlapping with each other. RESULTS: We applied our algorithm to several PPI (Protein-Protein Interaction) networks of Saccharomyces cerevisiae and Homo sapiens, and validated our results using public databases of protein complexes. The prediction accuracy was even more improved over our previous work which used also bottleneck information of the PPI network, but showed limitation when predicting small-sized protein complex detection. CONCLUSIONS: Our algorithm resulted in overlapping protein complexes with significantly improved F1 score over existing algorithms. This result comes from high recall due to effective network search, as well as high precision due to proper use of bottleneck information during the network search.