Detection of Alpha-Rod Protein Repeats Using a Neural Network and Application to Huntingtin

A growing number of solved protein structures display an elongated structural domain, denoted here as alpha-rod, composed of stacked pairs of anti-parallel alpha-helices. Alpha-rods are flexible and expose a large surface, which makes them suitable for protein interaction. Although most likely originating by tandem duplication of a two-helix unit, their detection using sequence similarity between repeats is poor. Here, we show that alpha-rod repeats can be detected using a neural network. The network detects more repeats than are identified by domain databases using multiple profiles, with a low level of false positives (<10%). We identify alpha-rod repeats in approximately 0.4% of proteins in eukaryotic genomes. We then investigate the results for all human proteins, identifying alpha-rod repeats for the first time in six protein families, including proteins STAG1-3, SERAC1, and PSMD1-2 & 5. We also characterize a short version of these repeats in eight protein families of Archaeal, Bacterial, and Fungal species. Finally, we demonstrate the utility of these predictions in directing experimental work to demarcate three alpha-rods in huntingtin, a protein mutated in Huntington''s disease. Using yeast two hybrid analysis and an immunoprecipitation technique, we show that the huntingtin fragments containing alpha-rods associate with each other. This is the first definition of domains in huntingtin and the first validation of predicted interactions between fragments of huntingtin, which sets up directions toward functional characterization of this protein. An implementation of the repeat detection algorithm is available as a Web server with a simple graphical output: http://www.ogic.ca/projects/ard. This can be further visualized using BiasViz, a graphic tool for representation of multiple sequence alignments.     

GRASP: Guided Reference-based Assembly of Short Peptides

Protein sequences predicted from metagenomic datasets are annotated by identifying their homologs via sequence comparisons with reference or curated proteins. However, a majority of metagenomic protein sequences are partial-length, arising as a result of identifying genes on sequencing reads or on assembled nucleotide contigs, which themselves are often very fragmented. The fragmented nature of metagenomic protein predictions adversely impacts homology detection and, therefore, the quality of the overall annotation of the dataset. Here we present a novel algorithm called GRASP that accurately identifies the homologs of a given reference protein sequence from a database consisting of partial-length metagenomic proteins. Our homology detection strategy is guided by the reference sequence, and involves the simultaneous search and assembly of overlapping database sequences. GRASP was compared to three commonly used protein sequence search programs (BLASTP, PSI-BLAST and FASTM). Our evaluations using several simulated and real datasets show that GRASP has a significantly higher sensitivity than these programs while maintaining a very high specificity. GRASP can be a very useful program for detecting and quantifying taxonomic and protein family abundances in metagenomic datasets. GRASP is implemented in GNU C++, and is freely available at http://sourceforge.net/projects/grasp-release.     

Pushing Structural Information into the Yeast Interactome by High-Throughput Protein Docking Experiments

The last several years have seen the consolidation of high-throughput proteomics initiatives to identify and characterize protein interactions and macromolecular complexes in model organisms. In particular, more that 10,000 high-confidence protein-protein interactions have been described between the roughly 6,000 proteins encoded in the budding yeast genome (Saccharomyces cerevisiae). However, unfortunately, high-resolution three-dimensional structures are only available for less than one hundred of these interacting pairs. Here, we expand this structural information on yeast protein interactions by running the first-ever high-throughput docking experiment with some of the best state-of-the-art methodologies, according to our benchmarks. To increase the coverage of the interaction space, we also explore the possibility of using homology models of varying quality in the docking experiments, instead of experimental structures, and assess how it would affect the global performance of the methods. In total, we have applied the docking procedure to 217 experimental structures and 1,023 homology models, providing putative structural models for over 3,000 protein-protein interactions in the yeast interactome. Finally, we analyze in detail the structural models obtained for the interaction between SAM1-anthranilate synthase complex and the MET30-RNA polymerase III to illustrate how our predictions can be straightforwardly used by the scientific community. The results of our experiment will be integrated into the general 3D-Repertoire pipeline, a European initiative to solve the structures of as many as possible protein complexes in yeast at the best possible resolution. All docking results are available at http://gatealoy.pcb.ub.es/HT_docking/.     

Identifying RNA-binding residues based on evolutionary conserved structural and energetic features

Increasing numbers of protein structures are solved each year, but many of these structures belong to proteins whose sequences are homologous to sequences in the Protein Data Bank. Nevertheless, the structures of homologous proteins belonging to the same family contain useful information because functionally important residues are expected to preserve physico-chemical, structural and energetic features. This information forms the basis of our method, which detects RNA-binding residues of a given RNA-binding protein as those residues that preserve physico-chemical, structural and energetic features in its homologs. Tests on 81 RNA-bound and 35 RNA-free protein structures showed that our method yields a higher fraction of true RNA-binding residues (higher precision) than two structure-based and two sequence-based machine-learning methods. Because the method requires no training data set and has no parameters, its precision does not degrade when applied to ‘novel’ protein sequences unlike methods that are parameterized for a given training data set. It was used to predict the ‘unknown’ RNA-binding residues in the C-terminal RNA-binding domain of human CPEB3. The two predicted residues, F430 and F474, were experimentally verified to bind RNA, in particular F430, whose mutation to alanine or asparagine nearly abolished RNA binding. The method has been implemented in a webserver called DR_bind1, which is freely available with no login requirement at http://drbind.limlab.ibms.sinica.edu.tw.     

Continuous Distributed Representation of Biological Sequences for Deep Proteomics and Genomics

We introduce a new representation and feature extraction method for biological sequences. Named bio-vectors (BioVec) to refer to biological sequences in general with protein-vectors (ProtVec) for proteins (amino-acid sequences) and gene-vectors (GeneVec) for gene sequences, this representation can be widely used in applications of deep learning in proteomics and genomics. In the present paper, we focus on protein-vectors that can be utilized in a wide array of bioinformatics investigations such as family classification, protein visualization, structure prediction, disordered protein identification, and protein-protein interaction prediction. In this method, we adopt artificial neural network approaches and represent a protein sequence with a single dense n-dimensional vector. To evaluate this method, we apply it in classification of 324,018 protein sequences obtained from Swiss-Prot belonging to 7,027 protein families, where an average family classification accuracy of 93%±0.06% is obtained, outperforming existing family classification methods. In addition, we use ProtVec representation to predict disordered proteins from structured proteins. Two databases of disordered sequences are used: the DisProt database as well as a database featuring the disordered regions of nucleoporins rich with phenylalanine-glycine repeats (FG-Nups). Using support vector machine classifiers, FG-Nup sequences are distinguished from structured protein sequences found in Protein Data Bank (PDB) with a 99.8% accuracy, and unstructured DisProt sequences are differentiated from structured DisProt sequences with 100.0% accuracy. These results indicate that by only providing sequence data for various proteins into this model, accurate information about protein structure can be determined. Importantly, this model needs to be trained only once and can then be applied to extract a comprehensive set of information regarding proteins of interest. Moreover, this representation can be considered as pre-training for various applications of deep learning in bioinformatics. The related data is available at Life Language Processing Website: http://llp.berkeley.edu and Harvard Dataverse: http://dx.doi.org/10.7910/DVN/JMFHTN.     

ProADD: A database on Protein Aggregation Diseases

ProADD, a database for protein aggregation diseases, is developed to organize the data under a single platform to facilitate easy access for researchers. Diseases caused due to protein aggregation and the proteins involved in each of these diseases are integrated. The database helps in classification of proteins involved in the protein aggregation diseases based on sequence and structural analysis. Analysis of proteins can be done to mine patterns prevailing among the aggregating proteins.

Availability

http://bicmku.in/ProADD     

Prediction and Analysis of Post-Translational Pyruvoyl Residue Modification Sites from Internal Serines in Proteins

Most of pyruvoyl-dependent proteins observed in prokaryotes and eukaryotes are critical regulatory enzymes, which are primary targets of inhibitors for anti-cancer and anti-parasitic therapy. These proteins undergo an autocatalytic, intramolecular self-cleavage reaction in which a covalently bound pyruvoyl group is generated on a conserved serine residue. Traditional detections of the modified serine sites are performed by experimental approaches, which are often labor-intensive and time-consuming. In this study, we initiated in an attempt for the computational predictions of such serine sites with Feature Selection based on a Random Forest. Since only a small number of experimentally verified pyruvoyl-modified proteins are collected in the protein database at its current version, we only used a small dataset in this study. After removing proteins with sequence identities >60%, a non-redundant dataset was generated and was used, which contained only 46 proteins, with one pyruvoyl serine site for each protein. Several types of features were considered in our method including PSSM conservation scores, disorders, secondary structures, solvent accessibilities, amino acid factors and amino acid occurrence frequencies. As a result, a pretty good performance was achieved in our dataset. The best 100.00% accuracy and 1.0000 MCC value were obtained from the training dataset, and 93.75% accuracy and 0.8441 MCC value from the testing dataset. The optimal feature set contained 9 features. Analysis of the optimal feature set indicated the important roles of some specific features in determining the pyruvoyl-group-serine sites, which were consistent with several results of earlier experimental studies. These selected features may shed some light on the in-depth understanding of the mechanism of the post-translational self-maturation process, providing guidelines for experimental validation. Future work should be made as more pyruvoyl-modified proteins are found and the method should be evaluated on larger datasets. At last, the predicting software can be downloaded from http://www.nkbiox.com/sub/pyrupred/index.html.     

GPS tracking data of Lesser Black-backed Gulls and Herring Gulls breeding at the southern North Sea coast

In this data paper, Bird tracking - GPS tracking of Lesser Black-backed Gulls and Herring Gulls breeding at the southern North Sea coast is described, a species occurrence dataset published by the Research Institute for Nature and Forest (INBO). The dataset (version 5.5) contains close to 2.5 million occurrences, recorded by 101 GPS trackers mounted on 75 Lesser Black-backed Gulls and 26 Herring Gulls breeding at the Belgian and Dutch coast. The trackers were developed by the University of Amsterdam Bird Tracking System (UvA-BiTS, http://www.uva-bits.nl). These automatically record and transmit bird movements, which allows us and others to study their habitat use and migration behaviour in great detail. Our bird tracking network is operational since 2013. It is funded for LifeWatch by the Hercules Foundation and maintained in collaboration with UvA-BiTS and the Flanders Marine Institute (VLIZ). The recorded data are periodically released in bulk as open data (http://dataset.inbo.be/bird-tracking-gull-occurrences), and are also accessible through CartoDB and the Global Biodiversity Information Facility (GBIF).     

FIGfams: yet another set of protein families

We present FIGfams, a new collection of over 100 000 protein families that are the product of manual curation and close strain comparison. Using the Subsystem approach the manual curation is carried out, ensuring a previously unattained degree of throughput and consistency. FIGfams are based on over 950 000 manually annotated proteins and across many hundred Bacteria and Archaea. Associated with each FIGfam is a two-tiered, rapid, accurate decision procedure to determine family membership for new proteins. FIGfams are freely available under an open source license. These can be downloaded at ftp://ftp.theseed.org/FIGfams/. The web site for FIGfams is http://www.theseed.org/wiki/FIGfams/     

An Ensemble Method for Predicting Subnuclear Localizations from Primary Protein Structures

Background

Predicting protein subnuclear localization is a challenging problem. Some previous works based on non-sequence information including Gene Ontology annotations and kernel fusion have respective limitations. The aim of this work is twofold: one is to propose a novel individual feature extraction method; another is to develop an ensemble method to improve prediction performance using comprehensive information represented in the form of high dimensional feature vector obtained by 11 feature extraction methods.

Methodology/Principal Findings

A novel two-stage multiclass support vector machine is proposed to predict protein subnuclear localizations. It only considers those feature extraction methods based on amino acid classifications and physicochemical properties. In order to speed up our system, an automatic search method for the kernel parameter is used. The prediction performance of our method is evaluated on four datasets: Lei dataset, multi-localization dataset, SNL9 dataset and a new independent dataset. The overall accuracy of prediction for 6 localizations on Lei dataset is 75.2% and that for 9 localizations on SNL9 dataset is 72.1% in the leave-one-out cross validation, 71.7% for the multi-localization dataset and 69.8% for the new independent dataset, respectively. Comparisons with those existing methods show that our method performs better for both single-localization and multi-localization proteins and achieves more balanced sensitivities and specificities on large-size and small-size subcellular localizations. The overall accuracy improvements are 4.0% and 4.7% for single-localization proteins and 6.5% for multi-localization proteins. The reliability and stability of our classification model are further confirmed by permutation analysis.

Conclusions

It can be concluded that our method is effective and valuable for predicting protein subnuclear localizations. A web server has been designed to implement the proposed method. It is freely available at http://bioinformatics.awowshop.com/snlpred_page.php.     

Building a Better Fragment Library for De Novo Protein Structure Prediction

Fragment-based approaches are the current standard for de novo protein structure prediction. These approaches rely on accurate and reliable fragment libraries to generate good structural models. In this work, we describe a novel method for structure fragment library generation and its application in fragment-based de novo protein structure prediction. The importance of correct testing procedures in assessing the quality of fragment libraries is demonstrated. In particular, the exclusion of homologs to the target from the libraries to correctly simulate a de novo protein structure prediction scenario, something which surprisingly is not always done. We demonstrate that fragments presenting different predominant predicted secondary structures should be treated differently during the fragment library generation step and that exhaustive and random search strategies should both be used. This information was used to develop a novel method, Flib. On a validation set of 41 structurally diverse proteins, Flib libraries presents both a higher precision and coverage than two of the state-of-the-art methods, NNMake and HHFrag. Flib also achieves better precision and coverage on the set of 275 protein domains used in the two previous experiments of the the Critical Assessment of Structure Prediction (CASP9 and CASP10). We compared Flib libraries against NNMake libraries in a structure prediction context. Of the 13 cases in which a correct answer was generated, Flib models were more accurate than NNMake models for 10. “Flib is available for download at: http://www.stats.ox.ac.uk/research/proteins/resources”.     

MRFalign: Protein Homology Detection through Alignment of Markov Random Fields

Sequence-based protein homology detection has been extensively studied and so far the most sensitive method is based upon comparison of protein sequence profiles, which are derived from multiple sequence alignment (MSA) of sequence homologs in a protein family. A sequence profile is usually represented as a position-specific scoring matrix (PSSM) or an HMM (Hidden Markov Model) and accordingly PSSM-PSSM or HMM-HMM comparison is used for homolog detection. This paper presents a new homology detection method MRFalign, consisting of three key components: 1) a Markov Random Fields (MRF) representation of a protein family; 2) a scoring function measuring similarity of two MRFs; and 3) an efficient ADMM (Alternating Direction Method of Multipliers) algorithm aligning two MRFs. Compared to HMM that can only model very short-range residue correlation, MRFs can model long-range residue interaction pattern and thus, encode information for the global 3D structure of a protein family. Consequently, MRF-MRF comparison for remote homology detection shall be much more sensitive than HMM-HMM or PSSM-PSSM comparison. Experiments confirm that MRFalign outperforms several popular HMM or PSSM-based methods in terms of both alignment accuracy and remote homology detection and that MRFalign works particularly well for mainly beta proteins. For example, tested on the benchmark SCOP40 (8353 proteins) for homology detection, PSSM-PSSM and HMM-HMM succeed on 48% and 52% of proteins, respectively, at superfamily level, and on 15% and 27% of proteins, respectively, at fold level. In contrast, MRFalign succeeds on 57.3% and 42.5% of proteins at superfamily and fold level, respectively. This study implies that long-range residue interaction patterns are very helpful for sequence-based homology detection. The software is available for download at http://raptorx.uchicago.edu/download/. A summary of this paper appears in the proceedings of the RECOMB 2014 conference, April 2–5.     

Engineering Proteins for Thermostability with iRDP Web Server

Engineering protein molecules with desired structure and biological functions has been an elusive goal. Development of industrially viable proteins with improved properties such as stability, catalytic activity and altered specificity by modifying the structure of an existing protein has widely been targeted through rational protein engineering. Although a range of factors contributing to thermal stability have been identified and widely researched, the in silico implementation of these as strategies directed towards enhancement of protein stability has not yet been explored extensively. A wide range of structural analysis tools is currently available for in silico protein engineering. However these tools concentrate on only a limited number of factors or individual protein structures, resulting in cumbersome and time-consuming analysis. The iRDP web server presented here provides a unified platform comprising of iCAPS, iStability and iMutants modules. Each module addresses different facets of effective rational engineering of proteins aiming towards enhanced stability. While iCAPS aids in selection of target protein based on factors contributing to structural stability, iStability uniquely offers in silico implementation of known thermostabilization strategies in proteins for identification and stability prediction of potential stabilizing mutation sites. iMutants aims to assess mutants based on changes in local interaction network and degree of residue conservation at the mutation sites. Each module was validated using an extensively diverse dataset. The server is freely accessible at http://irdp.ncl.res.in and has no login requirements.     

An Ensemble Method with Hybrid Features to Identify Extracellular Matrix Proteins

The extracellular matrix (ECM) is a dynamic composite of secreted proteins that play important roles in numerous biological processes such as tissue morphogenesis, differentiation and homeostasis. Furthermore, various diseases are caused by the dysfunction of ECM proteins. Therefore, identifying these important ECM proteins may assist in understanding related biological processes and drug development. In view of the serious imbalance in the training dataset, a Random Forest-based ensemble method with hybrid features is developed in this paper to identify ECM proteins. Hybrid features are employed by incorporating sequence composition, physicochemical properties, evolutionary and structural information. The Information Gain Ratio and Incremental Feature Selection (IGR-IFS) methods are adopted to select the optimal features. Finally, the resulting predictor termed IECMP (Identify ECM Proteins) achieves an balanced accuracy of 86.4% using the 10-fold cross-validation on the training dataset, which is much higher than results obtained by other methods (ECMPRED: 71.0%, ECMPP: 77.8%). Moreover, when tested on a common independent dataset, our method also achieves significantly improved performance over ECMPP and ECMPRED. These results indicate that IECMP is an effective method for ECM protein prediction, which has a more balanced prediction capability for positive and negative samples. It is anticipated that the proposed method will provide significant information to fully decipher the molecular mechanisms of ECM-related biological processes and discover candidate drug targets. For public access, we develop a user-friendly web server for ECM protein identification that is freely accessible at http://iecmp.weka.cc.     

mUbiSiDa: A Comprehensive Database for Protein Ubiquitination Sites in Mammals

Motivation

Protein ubiquitination is one of the important post-translational modifications by attaching ubiquitin to specific lysine (K) residues in target proteins, and plays important regulatory roles in many cell processes. Recent studies indicated that abnormal protein ubiquitination have been implicated in many diseases by degradation of many key regulatory proteins including tumor suppressor, oncoprotein, and cell cycle regulator. The detailed information of protein ubiquitination sites is useful for scientists to investigate the mechanism of many cell activities and related diseases.

Results

In this study we established mUbiSida for mammalian Ubiquitination Site Database, which provides a scientific community with a comprehensive, freely and high-quality accessible resource of mammalian protein ubiquitination sites. In mUbiSida, we deposited about 35,494 experimentally validated ubiquitinated proteins with 110,976 ubiquitination sites from five species. The mUbiSiDa can also provide blast function to predict novel protein ubiquitination sites in other species by blast the query sequence in the deposit sequences in mUbiSiDa. The mUbiSiDa was designed to be a widely used tool for biologists and biomedical researchers with a user-friendly interface, and facilitate the further research of protein ubiquitination, biological networks and functional proteomics. The mUbiSiDa database is freely available at http://reprod.njmu.edu.cn/mUbiSiDa.     

HIPPIE: Integrating protein interaction networks with experiment based quality scores

Protein function is often modulated by protein-protein interactions (PPIs) and therefore defining the partners of a protein helps to understand its activity. PPIs can be detected through different experimental approaches and are collected in several expert curated databases. These databases are used by researchers interested in examining detailed information on particular proteins. In many analyses the reliability of the characterization of the interactions becomes important and it might be necessary to select sets of PPIs of different confidence levels. To this goal, we generated HIPPIE (Human Integrated Protein-Protein Interaction rEference), a human PPI dataset with a normalized scoring scheme that integrates multiple experimental PPI datasets. HIPPIE's scoring scheme has been optimized by human experts and a computer algorithm to reflect the amount and quality of evidence for a given PPI and we show that these scores correlate to the quality of the experimental characterization. The HIPPIE web tool (available at http://cbdm.mdc-berlin.de/tools/hippie) allows researchers to do network analyses focused on likely true PPI sets by generating subnetworks around proteins of interest at a specified confidence level.     

Combining Physicochemical and Evolutionary Information for Protein Contact Prediction

We introduce a novel contact prediction method that achieves high prediction accuracy by combining evolutionary and physicochemical information about native contacts. We obtain evolutionary information from multiple-sequence alignments and physicochemical information from predicted ab initio protein structures. These structures represent low-energy states in an energy landscape and thus capture the physicochemical information encoded in the energy function. Such low-energy structures are likely to contain native contacts, even if their overall fold is not native. To differentiate native from non-native contacts in those structures, we develop a graph-based representation of the structural context of contacts. We then use this representation to train an support vector machine classifier to identify most likely native contacts in otherwise non-native structures. The resulting contact predictions are highly accurate. As a result of combining two sources of information—evolutionary and physicochemical—we maintain prediction accuracy even when only few sequence homologs are present. We show that the predicted contacts help to improve ab initio structure prediction. A web service is available at http://compbio.robotics.tu-berlin.de/epc-map/.     

A Model-Based Approach to Identify Binding Sites in CLIP-Seq Data

Cross-linking immunoprecipitation coupled with high-throughput sequencing (CLIP-Seq) has made it possible to identify the targeting sites of RNA-binding proteins in various cell culture systems and tissue types on a genome-wide scale. Here we present a novel model-based approach (MiClip) to identify high-confidence protein-RNA binding sites from CLIP-seq datasets. This approach assigns a probability score for each potential binding site to help prioritize subsequent validation experiments. The MiClip algorithm has been tested in both HITS-CLIP and PAR-CLIP datasets. In the HITS-CLIP dataset, the signal/noise ratios of miRNA seed motif enrichment produced by the MiClip approach are between 17% and 301% higher than those by the ad hoc method for the top 10 most enriched miRNAs. In the PAR-CLIP dataset, the MiClip approach can identify ∼50% more validated binding targets than the original ad hoc method and two recently published methods. To facilitate the application of the algorithm, we have released an R package, MiClip ( http://cran.r-project.org/web/packages/MiClip/index.html ), and a public web-based graphical user interface software (http://galaxy.qbrc.org/tool_runner?tool_id=mi_clip) for customized analysis.