Mining physical protein-protein interactions from the literature

Background:

Deciphering physical protein-protein interactions is fundamental to elucidating both the functions of proteins and biological processes. The development of high-throughput experimental technologies such as the yeast two-hybrid screening has produced an explosion in data relating to interactions. Since manual curation is intensive in terms of time and cost, there is an urgent need for text-mining tools to facilitate the extraction of such information. The BioCreative (Critical Assessment of Information Extraction systems in Biology) challenge evaluation provided common standards and shared evaluation criteria to enable comparisons among different approaches.

Results:

During the benchmark evaluation of BioCreative 2006, all of our results ranked in the top three places. In the task of filtering articles irrelevant to physical protein interactions, our method contributes a precision of 75.07%, a recall of 81.07%, and an AUC (area under the receiver operating characteristic curve) of 0.847. In the task of identifying protein mentions and normalizing mentions to molecule identifiers, our method is competitive among runs submitted, with a precision of 34.83%, a recall of 24.10%, and an F₁ score of28.5%. In extracting protein interaction pairs, our profile-based method was competitive on the SwissProt-only subset (precision = 36.95%, recall = 32.68%, and F₁ score = 30.40%) and on the entire dataset (30.96%, 29.35%, and26.20%, respectively). From the biologist's point of view, however, these findings are far from satisfactory. The error analysis presented in this report provides insight into how performance could be improved: three-quarters of false negatives were due to protein normalization problems (532/698), and about one-quarter were due to problems with correctly extracting interactions for this system.

Conclusion:

We present a text-mining framework to extract physical protein-protein interactions from the literature. Three key issues are addressed, namely filtering irrelevant articles, identifying protein names and normalizing them to molecule identifiers, and extracting protein-protein interactions. Our system is among the top three performers in the benchmark evaluation of BioCreative 2006. The tool will be helpful for manual interaction curation and can greatly facilitate the process of extracting protein-protein interactions.

     

Prediction of protein kinase-specific phosphorylation sites in hierarchical structure using functional information and random forest

Reversible protein phosphorylation is one of the most important post-translational modifications, which regulates various biological cellular processes. Identification of the kinase-specific phosphorylation sites is helpful for understanding the phosphorylation mechanism and regulation processes. Although a number of computational approaches have been developed, currently few studies are concerned about hierarchical structures of kinases, and most of the existing tools use only local sequence information to construct predictive models. In this work, we conduct a systematic and hierarchy-specific investigation of protein phosphorylation site prediction in which protein kinases are clustered into hierarchical structures with four levels including kinase, subfamily, family and group. To enhance phosphorylation site prediction at all hierarchical levels, functional information of proteins, including gene ontology (GO) and protein–protein interaction (PPI), is adopted in addition to primary sequence to construct prediction models based on random forest. Analysis of selected GO and PPI features shows that functional information is critical in determining protein phosphorylation sites for every hierarchical level. Furthermore, the prediction results of Phospho.ELM and additional testing dataset demonstrate that the proposed method remarkably outperforms existing phosphorylation prediction methods at all hierarchical levels. The proposed method is freely available at http://bioinformatics.ustc.edu.cn/phos_pred/.     

ConFunc--functional annotation in the twilight zone

Motivation: The success of genome sequencing has resulted inmany protein sequences without functional annotation. We presentConFunc, an automated Gene Ontology (GO)-based protein functionprediction approach, which uses conserved residues to generatesequence profiles to infer function. ConFunc split sets of sequencesidentified by PSI-BLAST into sub-alignments according to theirGO annotations. Conserved residues are identified for each GOterm sub-alignment for which a position specific scoring matrixis generated. This combination of steps produces a set of feature(GO annotation) derived profiles from which protein functionis predicted. Results: We assess the ability of ConFunc, BLAST and PSI-BLASTto predict protein function in the twilight zone of sequencesimilarity. ConFunc significantly outperforms BLAST & PSI-BLASTobtaining levels of recall and precision that are not obtainedby either method and maximum precision 24% greater than BLAST.Further for a large test set of sequences with homologues oflow sequence identity, at high levels of presicision, ConFuncobtains recall six times greater than BLAST. These results demonstratethe potential for ConFunc to form part of an automated genomicsannotation pipeline. Availability: http://www.sbg.bio.ic.ac.uk/confunc Contact: m.sternberg{at}imperial.ac.uk Supplementary information: Supplementary data are availableat Bioinformatics online. Associate Editor: Dmitrij Frishman     

AMS 4.0: consensus prediction of post-translational modifications in protein sequences

We present here the 2011 update of the AutoMotif Service (AMS 4.0) that predicts the wide selection of 88 different types of the single amino acid post-translational modifications (PTM) in protein sequences. The selection of experimentally confirmed modifications is acquired from the latest UniProt and Phospho.ELM databases for training. The sequence vicinity of each modified residue is represented using amino acids physico-chemical features encoded using high quality indices (HQI) obtaining by automatic clustering of known indices extracted from AAindex database. For each type of the numerical representation, the method builds the ensemble of Multi-Layer Perceptron (MLP) pattern classifiers, each optimising different objectives during the training (for example the recall, precision or area under the ROC curve (AUC)). The consensus is built using brainstorming technology, which combines multi-objective instances of machine learning algorithm, and the data fusion of different training objects representations, in order to boost the overall prediction accuracy of conserved short sequence motifs. The performance of AMS 4.0 is compared with the accuracy of previous versions, which were constructed using single machine learning methods (artificial neural networks, support vector machine). Our software improves the average AUC score of the earlier version by close to 7 % as calculated on the test datasets of all 88 PTM types. Moreover, for the selected most-difficult sequence motifs types it is able to improve the prediction performance by almost 32 %, when compared with previously used single machine learning methods. Summarising, the brainstorming consensus meta-learning methodology on the average boosts the AUC score up to around 89 %, averaged over all 88 PTM types. Detailed results for single machine learning methods and the consensus methodology are also provided, together with the comparison to previously published methods and state-of-the-art software tools. The source code and precompiled binaries of brainstorming tool are available at http://code.google.com/p/automotifserver/ under Apache 2.0 licensing.     

Prediction of S-Glutathionylation Sites Based on Protein Sequences

S-glutathionylation, the reversible formation of mixed disulfides between glutathione(GSH) and cysteine residues in proteins, is a specific form of post-translational modification that plays important roles in various biological processes, including signal transduction, redox homeostasis, and metabolism inside cells. Experimentally identifying S-glutathionylation sites is labor-intensive and time consuming, whereas bioinformatics methods provide an alternative way to this problem by predicting S-glutathionylation sites in silico. The bioinformatics approaches give not only candidate sites for further experimental verification but also bio-chemical insights into the mechanism of S-glutathionylation. In this paper, we firstly collect experimentally determined S-glutathionylated proteins and their corresponding modification sites from the literature, and then propose a new method for predicting S-glutathionylation sites by employing machine learning methods based on protein sequence data. Promising results are obtained by our method with an AUC (area under ROC curve) score of 0.879 in 5-fold cross-validation, which demonstrates the predictive power of our proposed method. The datasets used in this work are available at http://csb.shu.edu.cn/SGDB.     

Literature mining and database annotation of protein phosphorylation using a rule-based system

MOTIVATION: A large volume of experimental data on protein phosphorylation is buried in the fast-growing PubMed literature. While of great value, such information is limited in databases owing to the laborious process of literature-based curation. Computational literature mining holds promise to facilitate database curation. RESULTS: A rule-based system, RLIMS-P (Rule-based LIterature Mining System for Protein Phosphorylation), was used to extract protein phosphorylation information from MEDLINE abstracts. An annotation-tagged literature corpus developed at PIR was used to evaluate the system for finding phosphorylation papers and extracting phosphorylation objects (kinases, substrates and sites) from abstracts. RLIMS-P achieved a precision and recall of 91.4 and 96.4% for paper retrieval, and of 97.9 and 88.0% for extraction of substrates and sites. Coupling the high recall for paper retrieval and high precision for information extraction, RLIMS-P facilitates literature mining and database annotation of protein phosphorylation.     

Identification and characterization of the ATP·Mg-dependent protein phosphatase activator (F A) as a microtubule protein kinase in the brain

The activating factor of ATP·Mg-dependent protein phosphatase (F _A) has been identified in brain microtubules. When using purified MAP-2 (microtubule associated protein 2) and tau proteins as substrates,F _A could phosphorylate MAP-2 to 16 moles of phosphates per mole of protein with aK _m value of 0.4 µM, and tau proteins to 4 moles of phosphates per mole of proteins with aK _m value of about 3 µM. When using microtubules as substrates,F _A could enhance many-fold the endogenous phosphorylation of many microtubule-associated proteins including MAP-2, tau proteins, and several low-molecular-weight MAPs. In contrast to other reported MAP kinases, such as cAMP-dependent protein kinase and Ca⁺²/phospholipid-dependent protein kinase, theF _A-catalyzed phosphorylation of tau proteins could cause an electrophoretic mobility shift on sodium dodecyl sulfate polyacrylamide gel electrophoresis, suggesting that a dramatic conformational change of tau proteins was produced byF _A. Peptide mapping analysis of the phosphopeptides derived from SV8 protease digestion revealed thatF _A could phosphorylate MAP-2 and tau proteins on at least four specific sites distinctly different from those phosphorylated by cAMP-dependent and Ca⁺²/phospholipid-dependent MAP kinases. Quantitative analysis further indicated that approximately 19% of the total endogenous kinase activity in brain microtubules was due toF _A. Taken together, the results provide initial evidence that the ATP·Mg-dependent protein phosphatase activating factor (F _A) is a potent and unique MAP kinase, and may represent one of the major factors involved in phosphorylation of brain microtubules.     

Identifying protein phosphorylation sites with kinase substrate specificity on human viruses

Viruses infect humans and progress inside the body leading to various diseases and complications. The phosphorylation of viral proteins catalyzed by host kinases plays crucial regulatory roles in enhancing replication and inhibition of normal host-cell functions. Due to its biological importance, there is a desire to identify the protein phosphorylation sites on human viruses. However, the use of mass spectrometry-based experiments is proven to be expensive and labor-intensive. Furthermore, previous studies which have identified phosphorylation sites in human viruses do not include the investigation of the responsible kinases. Thus, we are motivated to propose a new method to identify protein phosphorylation sites with its kinase substrate specificity on human viruses. The experimentally verified phosphorylation data were extracted from virPTM - a database containing 301 experimentally verified phosphorylation data on 104 human kinase-phosphorylated virus proteins. In an attempt to investigate kinase substrate specificities in viral protein phosphorylation sites, maximal dependence decomposition (MDD) is employed to cluster a large set of phosphorylation data into subgroups containing significantly conserved motifs. The experimental human phosphorylation sites are collected from Phospho.ELM, grouped according to its kinase annotation, and compared with the virus MDD clusters. This investigation identifies human kinases such as CK2, PKB, CDK, and MAPK as potential kinases for catalyzing virus protein substrates as confirmed by published literature. Profile hidden Markov model is then applied to learn a predictive model for each subgroup. A five-fold cross validation evaluation on the MDD-clustered HMMs yields an average accuracy of 84.93% for Serine, and 78.05% for Threonine. Furthermore, an independent testing data collected from UniProtKB and Phospho.ELM is used to make a comparison of predictive performance on three popular kinase-specific phosphorylation site prediction tools. In the independent testing, the high sensitivity and specificity of the proposed method demonstrate the predictive effectiveness of the identified substrate motifs and the importance of investigating potential kinases for viral protein phosphorylation sites.     

Identifying human kinase-specific protein phosphorylation sites by integrating heterogeneous information from various sources

Phosphorylation is an important type of protein post-translational modification. Identification of possible phosphorylation sites of a protein is important for understanding its functions. Unbiased screening for phosphorylation sites by in vitro or in vivo experiments is time consuming and expensive; in silico prediction can provide functional candidates and help narrow down the experimental efforts. Most of the existing prediction algorithms take only the polypeptide sequence around the phosphorylation sites into consideration. However, protein phosphorylation is a very complex biological process in vivo. The polypeptide sequences around the potential sites are not sufficient to determine the phosphorylation status of those residues. In the current work, we integrated various data sources such as protein functional domains, protein subcellular location and protein-protein interactions, along with the polypeptide sequences to predict protein phosphorylation sites. The heterogeneous information significantly boosted the prediction accuracy for some kinase families. To demonstrate potential application of our method, we scanned a set of human proteins and predicted putative phosphorylation sites for Cyclin-dependent kinases, Casein kinase 2, Glycogen synthase kinase 3, Mitogen-activated protein kinases, protein kinase A, and protein kinase C families (available at http://cmbi.bjmu.edu.cn/huphospho). The predicted phosphorylation sites can serve as candidates for further experimental validation. Our strategy may also be applicable for the in silico identification of other post-translational modification substrates.     

Prediction of Protein Phosphorylation Sites by Using the Composition of k-Spaced Amino Acid Pairs

As one of the most widespread protein post-translational modifications, phosphorylation is involved in many biological processes such as cell cycle, apoptosis. Identification of phosphorylated substrates and their corresponding sites will facilitate the understanding of the molecular mechanism of phosphorylation. Comparing with the labor-intensive and time-consuming experiment approaches, computational prediction of phosphorylation sites is much desirable due to their convenience and fast speed. In this paper, a new bioinformatics tool named CKSAAP_PhSite was developed that ignored the kinase information and only used the primary sequence information to predict protein phosphorylation sites. The highlight of CKSAAP_PhSite was to utilize the composition of k-spaced amino acid pairs as the encoding scheme, and then the support vector machine was used as the predictor. The performance of CKSAAP_PhSite was measured with a sensitivity of 84.81%, a specificity of 86.07% and an accuracy of 85.43% for serine, a sensitivity of 78.59%, a specificity of 82.26% and an accuracy of 80.31% for threonine as well as a sensitivity of 74.44%, a specificity of 78.03% and an accuracy of 76.21% for tyrosine. Experimental results obtained from cross validation and independent benchmark suggested that our method was very promising to predict phosphorylation sites and can be served as a useful supplement tool to the community. For public access, CKSAAP_PhSite is available at http://59.73.198.144/cksaap_phsite/.     

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”.     

Tau protein kinase I/GSK-3Β/kinase FA in heparin phosphorylates tau on Ser199, Thr231, Ser235, Ser262, Ser369, and Ser400 sites phosphorylated in Alzheimer disease brain

Previously, tau protein kinase I/glycogen synthase kinase-3Β/kinase F_A(TPKI/GSK-3Β/F_A) was identified as a brain microtubule-associated tau kinase possibly involved in the Alzheimer disease-like phosphorylation of tau. In this report, we find that the TPKI/GSK-3Β/F_A can be stimulated to phosphorylate brain tau up to 8.5 mol of phosphates per mol of protein by heparin, a polyanion compound. Tryptic digestion of³²P-labeled tau followed by high-performance liquid chromatography and high-voltage electrophoresis/thin-layer chromatography reveals 12 phosphopeptides. Phosphoamino acid analysis together with sequential manual Edman degradation and peptide sequence analysis further reveals that TPKI/GSK-3/Β/F_A after heparin potentiation phosphorylates tau on sites of Ser¹⁹⁹, Thr²³¹, Ser²³⁵, Ser²⁶², Ser³⁹⁶, and Ser⁴⁰⁰, which are potential sites abnormally phosphorylated in Alzheimer tau and potent sites responsible for reducing microtubule binding possibly involved in neuronal degeneration. The results provide initial evidence that TPKI/GSK-3Β/F_A after heparin potentiation may represent one of the most potent systems possibly involved in the abnormal phosphorylation of PHF-tau and neuronal degeneration in Alzheimer disease brains.     

Beyond the clause: extraction of phosphorylation information from medline abstracts

MOTIVATION: Phosphorylation is an important biochemical reaction that plays a critical role in signal transduction pathways and cell-cycle processes. A text mining system to extract the phosphorylation relation from the literature is reported. The focus of this paper is on the new methods developed and implemented to connect and merge pieces of information about phosphorylation mentioned in different sentences in the text. The effectiveness and accuracy of the system as a whole as well as that of the methods for extraction beyond a clause/sentence is evaluated using an independently annotated dataset, the Phospho.ELM database. The new methods developed to merge pieces of information from different sentences are shown to be effective in significantly raising the recall without much difference in precision.     

hCKSAAP_UbSite: Improved prediction of human ubiquitination sites by exploiting amino acid pattern and properties

As one of the most common post-translational modifications, ubiquitination regulates the quantity and function of a variety of proteins. Experimental and clinical investigations have also suggested the crucial roles of ubiquitination in several human diseases. The complicated sequence context of human ubiquitination sites revealed by proteomic studies highlights the need of developing effective computational strategies to predict human ubiquitination sites. Here we report the establishment of a novel human-specific ubiquitination site predictor through the integration of multiple complementary classifiers. Firstly, a Support Vector Machine (SVM) classier was constructed based on the composition of k-spaced amino acid pairs (CKSAAP) encoding, which has been utilized in our previous yeast ubiquitination site predictor. To further exploit the pattern and properties of the ubiquitination sites and their flanking residues, three additional SVM classifiers were constructed using the binary amino acid encoding, the AAindex physicochemical property encoding and the protein aggregation propensity encoding, respectively. Through an integration that relied on logistic regression, the resulting predictor termed hCKSAAP_UbSite achieved an area under ROC curve (AUC) of 0.770 in 5-fold cross-validation test on a class-balanced training dataset. When tested on a class-balanced independent testing dataset that contains 3419 ubiquitination sites, hCKSAAP_UbSite has also achieved a robust performance with an AUC of 0.757. Specifically, it has consistently performed better than the predictor using the CKSAAP encoding alone and two other publicly available predictors which are not human-specific. Given its promising performance in our large-scale datasets, hCKSAAP_UbSite has been made publicly available at our server (http://protein.cau.edu.cn/cksaap_ubsite/).