Jaroslaw Meller

Combining prediction of secondary structure and solvent accessibility in proteins.

Owing to the use of evolutionary information and advanced machine learning protocols, secondary structures of amino acid residues in proteins can be predicted from the primary sequence with more than 75% per‐residue accuracy for the 3‐state (i.e., helix, β‐strand, and coil) classification problem. In this work we investigate whether further progress may be achieved by incorporating the relative solvent accessibility (RSA) of an amino acid residue as a fingerprint of the overall topology of the protein. Toward that goal, we developed a novel method for secondary structure prediction that uses predicted RSA in addition to attributes derived from evolutionary profiles. Our general approach follows the 2‐stage protocol of Rost and Sander, with a number of Elman‐type recurrent neural networks (NNs) combined into a consensus predictor. The RSA is predicted using our recently developed regression‐based method that provides real‐valued RSA, with the overall correlation coefficients between the actual and predicted RSA of about 0.66 in rigorous tests on independent control sets. Using the predicted RSA, we were able to improve the performance of our secondary structure prediction by up to 1.4% and achieved the overall per‐residue accuracy between 77.0% and 78.4% for the 3‐state classification problem on different control sets comprising, together, 603 proteins without homology to proteins included in the training. The effects of including solvent accessibility depend on the quality of RSA prediction. In the limit of perfect prediction (i.e., when using the actual RSA values derived from known protein structures), the accuracy of secondary structure prediction increases by up to 4%. We also observed that projecting real‐valued RSA into 2 discrete classes with the commonly used threshold of 25% RSA decreases the classification accuracy for secondary structure prediction. While the level of improvement of secondary structure prediction may be different for prediction protocols that implicitly account for RSA in other ways, we conclude that an increase in the 3‐state classification accuracy may be achieved when combining RSA with a state‐of‐the‐art protocol utilizing evolutionary profiles. The new method is available through a Web server at http://sable.cchmc.org. Proteins 2005.

Prediction‐based fingerprints of protein–protein interactions

The recognition of protein interaction sites is an important intermediate step toward identification of functionally relevant residues and understanding protein function, facilitating experimental efforts in that regard. Toward that goal, the authors propose a novel representation for the recognition of protein–protein interaction sites that integrates enhanced relative solvent accessibility (RSA) predictions with high resolution structural data. An observation that RSA predictions are biased toward the level of surface exposure consistent with protein complexes led the authors to investigate the difference between the predicted and actual (i.e., observed in an unbound structure) RSA of an amino acid residue as a fingerprint of interaction sites. The authors demonstrate that RSA prediction‐based fingerprints of protein interactions significantly improve the discrimination between interacting and noninteracting sites, compared with evolutionary conservation, physicochemical characteristics, structure‐derived and other features considered before. On the basis of these observations, the authors developed a new method for the prediction of protein–protein interaction sites, using machine learning approaches to combine the most informative features into the final predictor. For training and validation, the authors used several large sets of protein complexes and derived from them nonredundant representative chains, with interaction sites mapped from multiple complexes. Alternative machine learning techniques are used, including Support Vector Machines and Neural Networks, so as to evaluate the relative effects of the choice of a representation and a specific learning algorithm. The effects of induced fit and uncertainty of the negative (noninteracting) class assignment are also evaluated. Several representative methods from the literature are reimplemented to enable direct comparison of the results. Using rigorous validation protocols, the authors estimated that the new method yields the overall classification accuracy of about 74% and Matthews correlation coefficients of 0.42, as opposed to up to 70% classification accuracy and up to 0.3 Matthews correlation coefficient for methods that do not utilize RSA prediction‐based fingerprints. The new method is available at http://sppider.cchmc.org. Proteins 2007.

Accurate prediction of solvent accessibility using neural networks–based regression

Accurate prediction of relative solvent accessibilities (RSAs) of amino acid residues in proteins may be used to facilitate protein structure prediction and functional annotation. Toward that goal we developed a novel method for improved prediction of RSAs. Contrary to other machine learning–based methods from the literature, we do not impose a classification problem with arbitrary boundaries between the classes. Instead, we seek a continuous approximation of the real‐value RSA using nonlinear regression, with several feed forward and recurrent neural networks, which are then combined into a consensus predictor. A set of 860 protein structures derived from the PFAM database was used for training, whereas validation of the results was carefully performed on several nonredundant control sets comprising a total of 603 structures derived from new Protein Data Bank structures and had no homology to proteins included in the training. Two classes of alternative predictors were developed for comparison with the regression‐based approach: one based on the standard classification approach and the other based on a semicontinuous approximation with the so‐called thermometer encoding. Furthermore, a weighted approximation, with errors being scaled by the observed levels of variability in RSA for equivalent residues in families of homologous structures, was applied in order to improve the results. The effects of including evolutionary profiles and the growth of sequence databases were assessed. In accord with the observed levels of variability in RSA for different ranges of RSA values, the regression accuracy is higher for buried than for exposed residues, with overall 15.3–15.8% mean absolute errors and correlation coefficients between the predicted and experimental values of 0.64–0.67 on different control sets. The new method outperforms classification‐based algorithms when the real value predictions are projected onto two‐class classification problems with several commonly used thresholds to separate exposed and buried residues. For example, classification accuracy of about 77% is consistently achieved on all control sets with a threshold of 25% RSA. A web server that enables RSA prediction using the new method and provides customizable graphical representation of the results is available at http://sable.cchmc.org. Proteins 2004.

von Hippel-Lindau protein binds hyperphosphorylated large subunit of RNA polymerase II through a proline hydroxylation motif and targets it for ubiquitination

The transition from transcription initiation to elongation involves phosphorylation of the large subunit (Rpb1) of RNA polymerase II on the repetitive carboxyl-terminal domain. The elongating hyperphosphorylated Rpb1 is subject to ubiquitination, particularly in response to UV radiation and DNA-damaging agents. By using computer modeling, we identified regions of Rpb1 and the adjacent subunit 6 of RNA polymerase II (Rpb6) that share sequence and structural similarity with the domain of hypoxia-inducible transcription factor 1α (HIF-1α) that binds von Hippel–Lindau tumor suppressor protein (pVHL). pVHL confers substrate specificity to the E3 ligase complex, which ubiquitinates HIF-α and targets it for proteasomal degradation. In agreement with the computational model, we show biochemical evidence that pVHL specifically binds the hyperphosphorylated Rpb1 in a proline-hydroxylation-dependent manner, targeting it for ubiquitination. This interaction is regulated by UV radiation.

Mutations within the P2 Domain of Norovirus Capsid Affect Binding to Human Histo-Blood Group Antigens: Evidence for a Binding Pocket

ABSTRACT Noroviruses (NORs) are an important cause of acute gastroenteritis. Recent studies of NOR receptors showed that different NORs bind to different histo-blood group antigens (HBGAs), and at least four distinct binding patterns were observed. To determine the structure-function relationship for NORs and their receptors, two strains representing two of the four binding patterns were studied. Strain VA387 binds to HBGAs of A, B, and O secretors, whereas strain MOH binds to HBGAs of A and B secretors only. Using multiple sequence alignments, homology modeling, and structural analysis of NOR capsids, we identified a plausible “pocket” in the P2 domain that may be responsible for binding to HBGA receptors. This pocket consists of a conserved RGD/K motif surrounded by three strain-specific hot spots (N302, T337, and Q375 for VA387 and N302, N338, and E378 for MOH). Subsequent mutagenesis experiments demonstrated that all four sites played important roles in binding. A single amino acid mutation at T337 (to A) in VA387 or a double amino acid mutation at RN338 (to TT) in MOH abolished binding completely. Change of the entire RGD motif to SAS abolished binding in case of VA387, whereas single amino acid mutations in that motif did not have an apparent effect on binding to A and B antigens but decreased binding to H antigen. Multiple mutations at the RGK motif of MOH (SIRGK to TFRGD) completely knocked out the binding. Mutation of N302 or Q375 in VA387 affected binding to type O HBGA only, while switch mutants with three amino acid changes at either site from MOH to VA387 resulted in a weak binding to type O HBGAs. A further switch mutant with three amino acid changes at E378 from MOH to VA387 diminished the binding to type A HBGA only. Taken together, our data indicate that the binding pocket likely exists on NOR capsids. Direct evidence of this hypothesis requires crystallography studies.

Linear programming optimization and a double statistical filter for protein threading protocols.

The design of scoring functions (or potentials) for threading, differentiating native‐like from non‐native structures with a limited computational cost, is an active field of research. We revisit two widely used families of threading potentials: the pairwise and profile models. To design optimal scoring functions we use linear programming (LP). The LP protocol makes it possible to measure the difficulty of a particular training set in conjunction with a specific form of the scoring function. Gapless threading demonstrates that pair potentials have larger prediction capacity compared with profile energies. However, alignments with gaps are easier to compute with profile potentials. We therefore search and propose a new profile model with comparable prediction capacity to contact potentials. A protocol to determine optimal energy parameters for gaps, using LP, is also presented. A statistical test, based on a combination of local and global Z‐scores, is employed to filter out false‐positives. Extensive tests of the new protocol are presented. The new model provides an efficient alternative for threading with pair energies, maintaining comparable accuracy. The code, databases, and a prediction server are available at http://www.tc.cornell.edu/CBIO/loopp. Proteins 2001;45:241–261.

Cinteny: flexible analysis and visualization of synteny and genome rearrangements in multiple organisms

BackgroundIdentifying syntenic regions, i.e., blocks of genes or other markers with evolutionary conserved order, and quantifying evolutionary relatedness between genomes in terms of chromosomal rearrangements is one of the central goals in comparative genomics. However, the analysis of synteny and the resulting assessment of genome rearrangements are sensitive to the choice of a number of arbitrary parameters that affect the detection of synteny blocks. In particular, the choice of a set of markers and the effect of different aggregation strategies, which enable coarse graining of synteny blocks and exclusion of micro-rearrangements, need to be assessed. Therefore, existing tools and resources that facilitate identification, visualization and analysis of synteny need to be further improved to provide a flexible platform for such analysis, especially in the context of multiple genomes.ResultsWe present a new tool, Cinteny, for fast identification and analysis of synteny with different sets of markers and various levels of coarse graining of syntenic blocks. Using Hannenhalli-Pevzner approach and its extensions, Cinteny also enables interactive determination of evolutionary relationships between genomes in terms of the number of rearrangements (the reversal distance). In particular, Cinteny provides: i) integration of synteny browsing with assessment of evolutionary distances for multiple genomes; ii) flexibility to adjust the parameters and re-compute the results on-the-fly; iii) ability to work with user provided data, such as orthologous genes, sequence tags or other conserved markers. In addition, Cinteny provides many annotated mammalian, invertebrate and fungal genomes that are pre-loaded and available for analysis at http://cinteny.cchmc.org.ConclusionCinteny allows one to automatically compare multiple genomes and perform sensitivity analysis for synteny block detection and for the subsequent computation of reversal distances. Cinteny can also be used to interactively browse syntenic blocks conserved in multiple genomes, to facilitate genome annotation and validation of assemblies for newly sequenced genomes, and to construct and assess phylogenomic trees.

Versatile annotation and publication quality visualization of protein complexes using POLYVIEW-3D

BackgroundMacromolecular visualization as well as automated structural and functional annotation tools play an increasingly important role in the post-genomic era, contributing significantly towards the understanding of molecular systems and processes. For example, three dimensional (3D) models help in exploring protein active sites and functional hot spots that can be targeted in drug design. Automated annotation and visualization pipelines can also reveal other functionally important attributes of macromolecules. These goals are dependent on the availability of advanced tools that integrate better the existing databases, annotation servers and other resources with state-of-the-art rendering programs.ResultsWe present a new tool for protein structure analysis, with the focus on annotation and visualization of protein complexes, which is an extension of our previously developed POLYVIEW web server. By integrating the web technology with state-of-the-art software for macromolecular visualization, such as the PyMol program, POLYVIEW-3D enables combining versatile structural and functional annotations with a simple web-based interface for creating publication quality structure rendering, as well as animated images for Powerpoint™, web sites and other electronic resources. The service is platform independent and no plug-ins are required. Several examples of how POLYVIEW-3D can be used for structural and functional analysis in the context of protein-protein interactions are presented to illustrate the available annotation options.ConclusionPOLYVIEW-3D server features the PyMol image rendering that provides detailed and high quality presentation of macromolecular structures, with an easy to use web-based interface. POLYVIEW-3D also provides a wide array of options for automated structural and functional analysis of proteins and their complexes. Thus, the POLYVIEW-3D server may become an important resource for researches and educators in the fields of protein science and structural bioinformatics. The new server is available at http://polyview.cchmc.org/polyview3d.html.

Electronic excitation spectra of furan and pyrrole: Revisited by the symmetry adapted cluster-configuration interaction method

Electronic excitation spectra of furan and pyrrole are reinvestigated by the symmetry-adapted cluster configuration-interaction method. The 47 and 46 lowest singlet and triplet electronic states are computed for furan and pyrrole, respectively. Two series (1a2 and 2b1) of low-lying Rydberg states and the valence π–π* excited states strongly influence each other in both furan and pyrrole. The present calculations give detailed and satisfactory theoretical assignments of the vacuum ultraviolet spectra and the electron energy-loss spectra of the two molecules. The similarities and differences in the electronic excitations between furan and pyrrole are discussed in detail. The accuracy and assignments of recent theoretical studies, i.e., complete active space second-order perturbation, multireference Moller–Plesset perturbation, second-order algebraic-diagrammatic construction, multireference double configuration interaction, and CC3, are compared.

State‐specific coupled cluster‐type dressing of multireference singles and doubles configuration interaction matrix

Using the theory of state‐specific self‐consistent intermediate Hamiltonians, one proposes a new dressing of a multireference (MR) singles and doubles configuration interaction (CI) Hamiltonian matrix which insures size consistency. The method is based on a coupled cluster (CC) type factorization of the coefficients of the triples and quadruples and can be considered as leading to a dressed CI formulation of a state‐specific MRCC method. Preliminary application of the new procedure to the H4 model and comparison with other MRCC schemes are presented.