De novo protein structure prediction by incremental inter-residue geometries prediction and model quality assessment using deep learning

Motivation: The successful application of deep learning has promoted progress in protein model quality assessment. How to use model quality assessment to further improve the accuracy of protein structure prediction, especially not reliant on the existing templates, is helpful for unraveling the folding mechanism. Here, we investigate whether model quality assessment can be introduced into structure prediction to form a closed-loop feedback, and iteratively improve the accuracy of de novo protein structure prediction. Results: In this study, we propose a de novo protein structure prediction method called RocketX. In RocketX, a feedback mechanism is constructed through the geometric constraint prediction network GeomNet, the structural simulation module, and the model quality evaluation network EmaNet. In GeomNet, the co-evolutionary features extracted from MSA that search from the sequence databases are sent to an improved residual neural network to predict the inter-residue geometric constraints. The structure model is folded based on the predicted geometric constraints. In EmaNet, the 1D and 2D features are extracted from the folded model and sent to the deep residual neural network to estimate the inter-residue distance deviation and per-residue lDDT of the model, which will be fed back to GeomNet as dynamic features to correct the geometries prediction and progressively improve model accuracy. RocketX is tested on 483 benchmark proteins and 20 FM targets of CASP14. Experimental results show that the closed-loop feedback mechanism significantly contributes to the performance of RocketX, and the prediction accuracy of RocketX outperforms that of the state-of-the-art methods trRosetta (without templates) and RaptorX. In addition, the blind test results on CAMEO show that although no template is used, the prediction accuracy of RocketX on medium and hard targets is comparable to the advanced methods that integrate templates.

Superhydrophilicity of $\alpha$-Alumina Surfaces Results from Tight Binding of Interfacial Waters to Specific Aluminols

Understanding the microscopic driving force of water wetting is challenging and important for design of materials. In this work, we investigate, using classical molecular dynamics simulations, the water/$\alpha$-alumina (0001) and ($11\overline{2}0$) interfaces chosen for their chemical and physical differences. There is only one type of aluminol group on the nominally flat (0001) surface but three types on the microscopically rougher ($11\overline{2}0$) surface. We find that both surfaces are completely wet, consistent with contact angles of zero. Moreover, the work required to remove water from a nanoscale volume at the interface is larger for the (0001) surface than the ($11\overline{2}0$) surface, suggesting that the (0001) surface is more hydrophilic. In addition, translational and rotational dynamics of interfacial water molecules are slower than that in bulk water, suggesting tight binding to the surface. Interfacial waters show two major polar orientations, either pointing to or away from the solid surface. In the former case, waters donate strong hydrogen bonds to the surface, while in the latter they accept relatively weak ones from aluminol groups. The strength of hydrogen bonds is estimated using their lifetime and geometry. We found that for all aluminols, water-to-aluminol hydrogen bonds are stronger and have longer lifetimes than the aluminol-to-water ones. One exception is the long lifetime of the \ce{Al3OH}-water hydrogen bonds on the ($11\overline{2}0$) surface, due to geometric constraints. Interactions between surfaces and interfacial waters promote a templating effect whereby the latter are aligned in a pattern that follows the underlying lattice of the mineral surface.

Review of Subdivision Schemes and their Applications

Background: Subdivision surfaces modeling method and related technology research gradually become a hot spot in the field of computer-aided design(CAD) and computer graphics (CG). In the early stage, research on subdivision curves and surfaces mainly focused on the relationship between the points, thereby failing to satisfy the requirements of all geometric modeling. Considering many geometric constraints is necessary to construct subdivision curves and surfaces for achieving high-quality geometric modeling. Objective: This paper aims to summarize various subdivision schemes of subdivision curves and surfaces, particularly in geometric constraints, such as points and normals. The findings help scholars to grasp the current research status of subdivision curves and surfaces better and to explore their applications in geometric modeling. Methods: This paper reviews the theory and applications of subdivision schemes from four aspects. We first discuss the background and key concept of subdivision schemes. We then summarize the classification of classical subdivision schemes. Next, we show the subdivision surfaces fitting and summarize new subdivision schemes under geometric constraints. Applications of subdivision surfaces are also discussed. Finally, this paper gives a brief summary and future application prospects. Results: Many research papers and patents of subdivision schemes are classified in this review paper. Remarkable developments and improvements have been achieved in analytical computations and practical applications. Conclusion: Our review shows that subdivision curves and surfaces are widely used in geometric modeling. However, some topics need to be further studied. New subdivision schemes need to be presented to meet the requirements of new practical applications.

Monocular 3D Object Detection Based on Uncertainty Prediction of Keypoints

Three-dimensional (3D) object detection is an important task in the field of machine vision, in which the detection of 3D objects using monocular vision is even more challenging. We observe that most of the existing monocular methods focus on the design of the feature extraction framework or embedded geometric constraints, but ignore the possible errors in the intermediate process of the detection pipeline. These errors may be further amplified in the subsequent processes. After exploring the existing detection framework of keypoints, we find that the accuracy of keypoints prediction will seriously affect the solution of 3D object position. Therefore, we propose a novel keypoints uncertainty prediction network (KUP-Net) for monocular 3D object detection. In this work, we design an uncertainty prediction module to characterize the uncertainty that exists in keypoint prediction. Then, the uncertainty is used for joint optimization with object position. In addition, we adopt position-encoding to assist the uncertainty prediction, and use a timing coefficient to optimize the learning process. The experiments on our detector are conducted on the KITTI benchmark. For the two levels of easy and moderate, we achieve accuracy of 17.26 and 11.78 in AP3D, and achieve accuracy of 23.59 and 16.63 in APBEV, which are higher than the latest method KM3D.

Formation and dissociation of shear-induced high-energy dislocations: insight from molecular dynamics simulations

Solid-phase processing (SPP) allows one to create complex microstructures, not achievable via thermal processing alone. The resulting structures exhibit a rich palette of defects, both thermal and non-thermal, including defect substructures, such as dislocation networks. It is essential to understand the mechanisms of deformation and defect structure formation to guide SPP towards achieving desired microstructures and material properties. In this study, large-scale molecular dynamics simulations are used to investigate the effects of inhomogeneous strain distribution, that mimics deformation conditions of tribological tests, on the evolution of defects under severe shear deformation in polycrystalline Al. Analysis of defect nucleation and reaction pathways reveals that strong geometric constraints suppress the nucleation and slide of low energy dislocation 1/2<110>{111} but promote the nucleation and slide of high energy dislocations, such as [1-10](001) and 1/2[1-1-2](1-11). A rough contact surface, characteristic to tribological tests, imposes an inhomogeneous stress field leading to inhomogeneous defect substructures due to location-dependent activation of slip systems. The results suggest that high-energy dislocations can dominate the evolution of grain structures in highly constrained environments, which should be considered in modeling plastic deformation and grain refinement during SPP.

Development of an Experimental Setup to Investigate Influences on Component Distortion in Gravity Die Casting and a First Variation of Temperature Control Strategy

Distortion (1), residual stresses and hot cracks can facilitate significant decreases in quality characteristics of casting products. Their reduction by a suitable component design (2) and process control is therefore desirable. In the casting process, these characteristics are assumed as a result of the combination of solidification shrinkage paired with the local self-feeding and the geometric constraints imposed on the component by the mold. In gravity die casting (3) of aluminum (4) with thermally well conducting and rigid metal molds, the control of solidification through a localized adjustment of the heat balance (5) appears to be a suitable approach to minimize these effects. The development of an experimental setup for the assessment of the interdependencies of the alloy, casting geometry and cooling are described in this work. A first series of experiments with A356 aluminum alloy and the introduction to the different methods of evaluation are presented. Furthermore, an approach to improve the understanding of the underlying mechanisms is outlined.

User-Driven Computer-Assisted Reverse Engineering of Editable CAD Assembly Models

This paper introduces a novel reverse engineering technique for the reconstruction of editable CAD models of mechanical parts' assemblies. The input is a point cloud of a mechanical parts' assembly that has been acquired as a whole, i.e. without disassembling it prior to its digitization. The proposed framework allows for the reconstruction of the parametric CAD assembly model through a multi-step reconstruction and fitting approach. It is modular and it supports various exploitation scenarios depending on the available data and starting point. It also handles incomplete datasets. The reconstruction process starts from roughly sketched and parameterized geometries (i.e 2D sketches, 3D parts or assemblies) that are then used as input of a simulated annealing-based fitting algorithm, which minimizes the deviation between the point cloud and the reconstructed geometries. The coherence of the CAD models is maintained by a CAD modeler that performs the updates and satisfies the geometric constraints as the fitting process goes on. The optimization process leverages a two-level filtering technique able to capture and manage the boundaries of the geometries inside the overall point cloud in order to allow for local fitting and interfaces detection. It is a user-driven approach where the user decides what are the most suitable steps and sequence to operate. It has been tested and validated on both real scanned point clouds and as-scanned virtually generated point clouds incorporating several artifacts that would appear with real acquisition devices.

Use of instrumental variables methods in trajectory measurements’ problems

Possibilities of theinstrumental variables method using to improve the matrix properties of algebraic equations system are shown in the article. The problem of trajectory measurements from a network of ground-based, no-demand measuring stations using GLONASS navigation satellites is considered. The need to increase information content of trajectory measurements is especially important for trajectory measurements in the southern hemisphere of the Earth with a small number of measuring stations. The numerical experiment results with the simulation modeling use of trajectory measurements of the navigation satellite movement from two and three request-free measuring stations with geometric constraints are presented. The possibility of the measurements’ information content increasing from such base stations using the instrumental variables method is shown. Comparison of an indicator of the trajectory measurements informativenessis made according to the degree of the matrix conditionality of the algebraic equations system being solved. Based on the numerical experiment results, it isconsidered thepossibility to increase the information content of trajectory measurements from a network of non-query measuring stations with geometric constraints, with the instrumental variables involvement in the method processing. A quantitative estimate of the increase in information content was obtained in the form of a decrease in the condition number of the system matrix being solved by more than one order of magnitude. In the future, the research is predicted on the possibility of the trajectory measurements informativeness increasing for various basic functions used as instrumental variables, and their effect on the geometric constraints reducing an uneven network of measuring stations.

Electron Density Geometry and the Quantum Theory of Atoms in Molecules

A novel form of charge density analysis, that of isosurface curvature redistribution, is formulated and applied to the toy problem of carbonyl oxygen activation in formaldehyde. The isosurface representation of the electron charge density allows us to incorporate the rigorous geometric constraints of closed surfaces towards the analysis and chemical interpretation of the charge density response to perturbations. Visual inspection of 2D isosurface motion resulting from applied external electric fields reveals how isosurface curvature flows within and between atoms, and that a molecule can be uniquely and completely partitioned into chemically significant regions of positive and negative curvature. These concepts reveal that carbonyl oxygen activation proceeds primarily through curvature and charge redistribution within rather than between Bader atoms. Using gradient bundle analysis—the partitioning of formaldehyde into infinitesimal volume elements bounded by QTAIM zero flux surfaces—the observations from visual isosurface inspection are verified. The results of the formaldehyde carbonyl analysis are then shown to be transferable to the substrate carbonyl in the ketosteroid isomerase enzyme, laying the groundwork for extending this approach to the problems of enzymatic catalysis.