Residue-level Model Research Articles

Assessing protein model quality based on deep graph coupled networks using protein language model.

Model quality evaluation is a crucial part of protein structural biology. How to distinguish high-quality models from low-quality models, and to assess which high-quality models have relatively incorrect regions for improvement, are remain a challenge. More importantly, the quality assessment of multimer models is a hot topic for structure prediction. In this study, we propose GraphCPLMQA, a novel approach for evaluating residue-level model quality that combines graph coupled networks and embeddings from protein language models. The GraphCPLMQA consists of a graph encoding module and a transform-based convolutional decoding module. In encoding module, the underlying relational representations of sequence and high-dimensional geometry structure are extracted by protein language models with Evolutionary Scale Modeling. In decoding module, the mapping connection between structure and quality is inferred by the representations and low-dimensional features. Specifically, the triangular location and residue level contact order features are designed to enhance the association between the local structure and the overall topology. Experimental results demonstrate that GraphCPLMQA using single-sequence embedding achieves the best performance compared with the CASP15 residue-level interface evaluation methods among 9108 models in the local residue interface test set of CASP15 multimers. In CAMEO blind test (20 May 2022 to 13 August 2022), GraphCPLMQA ranked first compared with other servers (https://www.cameo3d.org/quality-estimation). GraphCPLMQA also outperforms state-of-the-art methods on 19,035 models in CASP13 and CASP14 monomer test set.

DeepCellEss: cell line-specific essential protein prediction with attention-based interpretable deep learning.

Protein essentiality is usually accepted to be a conditional trait and strongly affected by cellular environments. However, existing computational methods often do not take such characteristics into account, preferring to incorporate all available data and train a general model for all cell lines. In addition, the lack of model interpretability limits further exploration and analysis of essential protein predictions. In this study, we proposed DeepCellEss, a sequence-based interpretable deep learning framework for cell line-specific essential protein predictions. DeepCellEss utilizes a convolutional neural network and bidirectional long short-term memory to learn short- and long-range latent information from protein sequences. Further, a multi-head self-attention mechanism is used to provide residue-level model interpretability. For model construction, we collected extremely large-scale benchmark datasets across 323 cell lines. Extensive computational experiments demonstrate that DeepCellEss yields effective prediction performance for different cell lines and outperforms existing sequence-based methods as well as network-based centrality measures. Finally, we conducted some case studies to illustrate the necessity of considering specific cell lines and the superiority of DeepCellEss. We believe that DeepCellEss can serve as a useful tool for predicting essential proteins across different cell lines. The DeepCellEss web server is available at http://csuligroup.com:8000/DeepCellEss. The source code and data underlying this study can be obtained from https://github.com/CSUBioGroup/DeepCellEss. Supplementary data are available at Bioinformatics online.

Exploring allosteric communication in multiple states of the bacterial ribosome using residue network analysis.

Antibiotic resistance is one of the most important problems of our era and hence the discovery of new effective therapeutics is urgent. At this point, studying the allosteric communication pathways in the bacterial ribosome and revealing allosteric sites/residues is critical for designing new inhibitors or repurposing readily approved drugs for this enormous machine. To shed light onto molecular details of the allosteric mechanisms, here we construct residue networks of the bacterial ribosomal complex at four different states of translation by using an effective description of the intermolecular interactions. Centrality analysis of these networks highlights the functional roles of structural components and critical residues on the ribosomal complex. High betweenness scores reveal pathways of residues connecting numerous sites on the structure. Interestingly, these pathways assemble highly conserved residues, drug binding sites, and known allosterically linked regions on the same structure. This study proposes a new residue-level model to test how distant sites on the molecular machine may be linked through hub residues that are critically located on the contact topology to inherently form communication pathways. Findings also indicate intersubunit bridges B1b, B3, B5, B7, and B8 as critical targets to design novel antibiotics.

Discrete structural features among interface residue-level classes.

BackgroundProtein-protein interaction (PPI) is essential for molecular functions in biological cells. Investigation on protein interfaces of known complexes is an important step towards deciphering the driving forces of PPIs. Each PPI complex is specific, sensitive and selective to binding. Therefore, we have estimated the relative difference in percentage of polar residues between surface and the interface for each complex in a non-redundant heterodimer dataset of 278 complexes to understand the predominant forces driving binding.ResultsOur analysis showed ~60% of protein complexes with surface polarity greater than interface polarity (designated as class A). However, a considerable number of complexes (~40%) have interface polarity greater than surface polarity, (designated as class B), with a significantly different p-value of 1.66E-45 from class A. Comprehensive analyses of protein complexes show that interface features such as interface area, interface polarity abundance, solvation free energy gain upon interface formation, binding energy and the percentage of interface charged residue abundance distinguish among class A and class B complexes, while electrostatic visualization maps also help differentiate interface classes among complexes.ConclusionsClass A complexes are classical with abundant non-polar interactions at the interface; however class B complexes have abundant polar interactions at the interface, similar to protein surface characteristics. Five physicochemical interface features analyzed from the protein heterodimer dataset are discriminatory among the interface residue-level classes. These novel observations find application in developing residue-level models for protein-protein binding prediction, protein-protein docking studies and interface inhibitor design as drugs.

ResQ: An Approach to Unified Estimation of B-Factor and Residue-Specific Error in Protein Structure Prediction

Computer-based structure prediction becomes a major tool to provide large-scale structure models for annotating biological function of proteins. Information of residue-level accuracy and thermal mobility (or B-factor), which is critical to decide how biologists utilize the predicted models, is however missed in most structure prediction pipelines. We developed ResQ for unified residue-level model quality and B-factor estimations by combining local structure assembly variations with sequence-based and structure-based profiling. ResQ was tested on 635 non-redundant proteins with structure models generated by I-TASSER, where the average difference between estimated and observed distance errors is 1.4Å for the confidently modeled proteins. ResQ was further tested on structure decoys from CASP9-11 experiments, where the error of local structure quality prediction is consistently lower than or comparable to other state-of-the-art predictors. Finally, ResQ B-factor profile was used to assist molecular replacement, which resulted in successful solutions on several proteins that could not be solved from constant B-factor settings.

Molecule-Centered Method for Accelerating the Calculation of Hydrodynamic Interactions in Brownian Dynamics Simulations Containing Many Flexible Biomolecules

Inclusion of hydrodynamic interactions (HIs) is essential in simulations of biological macromolecules that treat the solvent implicitly if the macromolecules are to exhibit correct translational and rotational diffusion. The present work describes the development and testing of a simple approach aimed at allowing more rapid computation of HIs in coarse-grained Brownian dynamics simulations of systems that contain large numbers of flexible macromolecules. The method combines a complete treatment of intramolecular HIs with an approximate treatment of the intermolecular HIs which assumes that the molecules are effectively spherical; all of the HIs are calculated at the Rotne-Prager-Yamakawa level of theory. When combined with Fixman's Chebyshev polynomial method for calculating correlated random displacements, the proposed method provides an approach that is simple to program but sufficiently fast that it makes it computationally viable to include HIs in large-scale simulations. Test calculations performed on very coarse-grained models of the pyruvate dehydrogenase (PDH) E2 complex and on oligomers of ParM (ranging in size from 1 to 20 monomers) indicate that the method reproduces the translational diffusion behavior seen in more complete HI simulations surprisingly well; the method performs less well at capturing rotational diffusion but its discrepancies diminish with increasing size of the simulated assembly. Simulations of residue-level models of two tetrameric protein models demonstrate that the method also works well when more structurally detailed models are used in the simulations. Finally, test simulations of systems containing up to 1024 coarse-grained PDH molecules indicate that the proposed method rapidly becomes more efficient than the conventional BD approach in which correlated random displacements are obtained via a Cholesky decomposition of the complete diffusion tensor.

Prediction of Hydrodynamic and Other Solution Properties of Partially Disordered Proteins with a Simple, Coarse-Grained Model.

The possibility of validating structures of intrinsically disordered proteins against solution properties is a goal that would be most helpful in the understanding of their function. We have devised a scheme for the prediction of solution properties of partially disordered proteins that comprise one or more ordered domains, along with flexible tails or linkers. A very simple, coarse-grained, residue-level model, which is easily parametrized using available structural information, along with previously developed tools for the simulation of solution conformation and dynamics, allows the prediction of properties like sedimentation coefficients, relaxation times, and X-ray or neutron scattering. This is demonstrated for a variety of partially disordered proteins, for which well-characterized solution properties are very accurately evaluated, with predictions falling in most cases within experimental errors.

Prediction of Hydrodynamic and Other Solution Properties of Rigid Proteins from Atomic- and Residue-Level Models

Here we extend the ability to predict hydrodynamic coefficients and other solution properties of rigid macromolecular structures from atomic-level structures, implemented in the computer program HYDROPRO, to models with lower, residue-level resolution. Whereas in the former case there is one bead per nonhydrogen atom, the latter contains one bead per amino acid (or nucleotide) residue, thus allowing calculations when atomic resolution is not available or coarse-grained models are preferred. We parameterized the effective hydrodynamic radius of the elements in the atomic- and residue-level models using a very large set of experimental data for translational and rotational coefficients (intrinsic viscosity and radius of gyration) for >50 proteins. We also extended the calculations to very large proteins and macromolecular complexes, such as the whole 70S ribosome. We show that with proper parameterization, the two levels of resolution yield similar and rather good agreement with experimental data. The new version of HYDROPRO, in addition to considering various computational and modeling schemes, is far more efficient computationally and can be handled with the use of a graphical interface.

Calculations of the binding affinities of protein-protein complexes with the fast multipole method

In this paper, we used a coarse-grained model at the residue level to calculate the binding free energies of three protein-protein complexes. General formulations to calculate the electrostatic binding free energy and the van der Waals free energy are presented by solving linearized Poisson-Boltzmann equations using the boundary element method in combination with the fast multipole method. The residue level model with the fast multipole method allows us to efficiently investigate how the mutations on the active site of the protein-protein interface affect the changes in binding affinities of protein complexes. Good correlations between the calculated results and the experimental ones indicate that our model can capture the dominant contributions to the protein-protein interactions. At the same time, additional effects on protein binding due to atomic details are also discussed in the context of the limitations of such a coarse-grained model.

Protein flexibility prediction by an all‐atom mean‐field statistical theory

We extended a mean-field model to proteins with all atomic detail. The all-atom mean-field model was used to calculate the dynamic and thermodynamic properties of a three-helix bundle fragment of Staphylococcal protein A (Protein Data Bank [PDB] ID 1BDD) and alpha-spectrin SH3 domain protein (PDB ID 1SHG). We show that a model with all-atomic detail provides a significantly more accurate prediction of flexibility of residues in proteins than does a coarse-grained residue-level model. The accuracy of flexibility prediction is further confirmed by application of the method to 18 additional proteins with the largest size of 224 residues.