Pose Prediction Research Articles

Accurate prediction of protein–ligand interactions by combining physical energy functions and graph-neural networks

We introduce an advanced model for predicting protein–ligand interactions. Our approach combines the strengths of graph neural networks with physics-based scoring methods. Existing structure-based machine-learning models for protein–ligand binding prediction often fall short in practical virtual screening scenarios, hindered by the intricacies of binding poses, the chemical diversity of drug-like molecules, and the scarcity of crystallographic data for protein–ligand complexes. To overcome the limitations of existing machine learning-based prediction models, we propose a novel approach that fuses three independent neural network models. One classification model is designed to perform binary prediction of a given protein–ligand complex pose. The other two regression models are trained to predict the binding affinity and root-mean-square deviation of a ligand conformation from an input complex structure. We trained the model to account for both deviations in experimental and predicted binding affinities and pose prediction uncertainties. By effectively integrating the outputs of the triplet neural networks with a physics-based scoring function, our model showed a significantly improved performance in hit identification. The benchmark results with three independent decoy sets demonstrate that our model outperformed existing models in forward screening. Our model achieved top 1% enrichment factors of 32.7 and 23.1 with the CASF2016 and DUD-E benchmark sets, respectively. The benchmark results using the LIT-PCBA set further confirmed its higher average enrichment factors, emphasizing the model’s efficiency and generalizability. The model’s efficiency was further validated by identifying 23 active compounds from 63 candidates in experimental screening for autotaxin inhibitors, demonstrating its practical applicability in hit discovery.Scientific contributionOur work introduces a novel training strategy for a protein–ligand binding affinity prediction model by integrating the outputs of three independent sub-models and utilizing expertly crafted decoy sets. The model showcases exceptional performance across multiple benchmarks. The high enrichment factors in the LIT-PCBA benchmark demonstrate its potential to accelerate hit discovery.

Determining the Optimal Tradeoff between Compute and Accuracy for Diffusion Models in Molecular Docking

In the world of molecular biology, Molecular Docking has become a pillar for understanding complex interactions pivotal for drug design and more. The emergence of sophisticated machine learning models, such as Diffusion Generative Models, have significantly expanded our analytical capabilities. However, this evolution has introduced notable computational challenges. This study aims to examine the trade-off between computational demands and model accuracy. Our methodology, which incorporates free platforms, illuminates methods to conserve computational resources while maintaining nearoptimal accuracy. Our findings suggest that using 30 samples per complex, 15 inference steps, and 4 batch steps improves pose prediction accuracy and reduces computational resources. The proposed parameters achieve a 14% accuracy increase compared to the 40 samples per complex model and a 56.25% increase compared to the 10 samples per complex model. The optimized inference steps result in a 12.2% accuracy increase over the 20-step control using the 40 samples model and a 24.3% increase using the 10 samples model. Additionally, using 4 batch steps leads to a 40.6% increase in DiffDock Confidence for the 10-sample control and a 0.4% increase for the 40- sample control.

Improving docking and virtual screening performance using AlphaFold2 multi-state modeling for kinases

Structure-based virtual screening (SBVS) is a crucial computational approach in drug discovery, but its performance is sensitive to structural variations. Kinases, which are major drug targets, exemplify this challenge due to active site conformational changes caused by different inhibitor types. Most experimentally determined kinase structures have the DFGin state, potentially biasing SBVS towards type I inhibitors and limiting the discovery of diverse scaffolds. We introduce a multi-state modeling (MSM) protocol for AlphaFold2 (AF2) kinase structures using state-specific templates to address these challenges. Our comprehensive benchmarks evaluate predicted model qualities, binding pose prediction accuracy, and hit compound identification through ensemble SBVS. Results demonstrate that MSM models exhibit comparable or improved structural accuracy compared to standard AF2 models, enhancing pose prediction accuracy and effectively capturing kinase-ligand interactions. In virtual screening experiments, our MSM approach consistently outperforms standard AF2 and AF3 modeling, particularly in identifying diverse hit compounds. This study highlights the potential of MSM in broadening kinase inhibitor discovery by facilitating the identification of chemically diverse inhibitors, offering a promising solution to the structural bias problem in kinase-targeted drug discovery.

Structure-based pose prediction: Non-cognate docking extended to macrocyclic ligands

So-called “cross-docking” is the prediction of the bound configuration of small-molecule ligands that differ from the cognate ligand of a protein co-crystal structure. This is a much more challenging problem than re-docking the cognate ligand, particularly when the new ligand is structurally dissimilar from prior known ones. We have updated the previously introduced PINC (“PINC Is Not Cognate”) benchmark which introduced the idea of temporal segregation to measure cross-docking performance. The temporal set encompasses 846 future ligands for ten targets based on information from the earliest 25% of X-ray co-crystal structures known for each target. Here, we extend the benchmark to include thirteen targets where the bound poses of 128 macrocyclic ligands are to be predicted based on knowledge from structures of bound non-macrocyclic ligands. Performance was roughly equivalent for both the temporally-split non-macrocyclic ligand set and the macrocycle prediction set. Using standard and fully automatic protocols for the Surflex-Dock and ForceGen methods, across the combined 974 non-macrocyclic and macrocyclic ligands, the top-scoring pose family was correct 68% of the time, with the top-two pose families achieving a 79% success rate. Correct poses among all those predicted were identified 92% of the time. These success rates far exceeded those observed for the alternative methods AutoDock Vina and Gnina on both sets.

DM-HAP: Diffusion model for accurate hand pose prediction

Forecasting hand poses is a challenging task due to inherent uncertainties, occlusions, and inaccuracies in 3D pose estimation. Diffusion models provide a promising direction for predicting precise 3D hand poses under noisy conditions. In this work, we introduce Dual-diffusion, an innovative framework for the precise prediction of future hand poses. Our approach leverages the strengths of diffusion models by framing hand pose forecasting as a reverse diffusion process, effectively addressing the complexities of noisy hand joint movements and their subtleties. To enhance the learning of hand pose representations, we propose a unique neural architecture that simultaneously captures both local and global features. This is achieved through the deployment of Global and Local Diffusion (GLD) blocks within our network, which facilitate the exchange of information between local and global features. This diffusion-based interaction enables the integration of global hand actions and local finger actions, leading to a more powerful representation learning approach. We evaluate the effectiveness of our Dual-diffusion method on three public 3D hand pose estimation datasets (MSRA, F-PHAB, and BigHand2.2M), and our approach outperforms previous methods.

YOLOv5-POS: research on cabbage pose prediction method based on multi-task perception technology.

Accurate and rapid identification of cabbage posture is crucial for minimizing damage to cabbage heads during mechanical harvesting. However, due to the structural complexity of cabbages, current methods encounter challenges in detecting and segmenting the heads and roots. Therefore, exploring efficient cabbage posture prediction methods is of great significance. This study introduces YOLOv5-POS, an innovative cabbage posture prediction approach. Building on the YOLOv5s backbone, this method enhances detection and segmentation capabilities for cabbage heads and roots by incorporating C-RepGFPN to replace the traditional Neck layer, optimizing feature extraction and upsampling strategies, and refining the C-Seg segmentation head. Additionally, a cabbage root growth prediction model based on Bézier curves is proposed, using the geometric moment method for key point identification and the anti-gravity stem-seeking principle to determine root-head junctions. It performs precision root growth curve fitting and prediction, effectively overcoming the challenge posed by the outer leaves completely enclosing the cabbage root stem. YOLOv5-POS was tested on a multi-variety cabbage dataset, achieving an F1 score of 98.8% for head and root detection, with an instance segmentation accuracy of 93.5%. The posture recognition model demonstrated an average absolute error of 1.38° and an average relative error of 2.32%, while the root growth prediction model reached an accuracy of 98%. Cabbage posture recognition was completed within 28 milliseconds, enabling real-time harvesting. The enhanced model effectively addresses the challenges of cabbage segmentation and posture prediction, providing a highly accurate and efficient solution for automated harvesting, minimizing crop damage, and improving operational efficiency.

Sparse Convolution-Based 6D Pose Estimation for Robotic Bin-Picking With Point Clouds

Abstract Estimating the orientation and position of objects is a crucial step in robotic bin-picking tasks. The challenge lies in the fact that, in real-world scenarios, a diverse array of objects is often randomly stacked, resulting in significant occlusion. This study introduces an innovative approach aimed at predicting 6D poses by processing point clouds through a two-stage neural network. In the initial stage, a network for scenes with low-textured environments is designed. Its purpose is to perform instance segmentation and provide an initial pose estimation. Entering the second stage, a pose refinement network is suggested. This network is intended to enhance the precision of pose prediction, building upon the output from the first stage. To tackle the challenge of resource-intensive annotation, a simulation technique is employed to generate a synthetic dataset. Additionally, a dedicated software tool has been developed to annotate real point cloud datasets. In practical experiments, our method demonstrated superior performance compared to baseline methods such as PointGroup and Iterative Closest Point. This superiority is evident in both segmentation accuracy and pose refinement. Moreover, practical grasping experiments have underscored the method's efficacy in real-world industrial robot bin-picking applications. The results affirm its capability to successfully address the challenges produced by occluded and randomly stacked objects.

Development of Drug Discovery Platforms Using Artificial Intelligence and Cheminformatics.

Recently, remarkable progress has been achieved in artificial intelligence (AI), including machine learning. Various AI models have been proposed for drug discovery, including the design of small molecules, activity prediction, and three-dimensional (3D) structure prediction of proteins. AI consists of diverse elements, including information retrieval and machine learning, and can be used in a wide range of drug discovery scenarios. In this review, we focused on AI for small-molecule drug discovery with respect to molecular design, activity prediction, and prediction of the binding poses of compounds to target molecules. We also discussed the applications of AI in academic drug discovery.

Shield machine pose prediction based on CNN-GRU-Attention

This paper presents a shield machine pose prediction method based on Convolutional Neural Network (CNN) - Gated Recurrent Unit (GRU) with Attention mechanism (Attention). Firstly, the Pearson correlation coefficient is employed to select input parameters highly related to the position and posture of the shield machine. Then, a convolutional neural network is introduced to extract the long-term short-term feature dependency features in the operation data of the shield machine, optimizing the model’s input. The attention mechanism is integrated into the gated loop unit to make the model more targeted in using key information in the input sequence and improve the accuracy of the shield machine pose prediction model. The effectiveness of this method is verified by the example of Beijing Metro Line 10. Compared with GRU-Attention and LSTM-Attention models, the mean value of determination coefficient R2 increased from 0.872 and 0.886 to 0.959, and the mean value of root mean square error RMSE decreased from 2.78 and 2.52 to 2.14. This method can provide effective prediction for the attitude and position of shield machines in actual tunnel engineering.

ESMformer: Error-aware self-supervised transformer for multi-view 3D human pose estimation

Multi-view 3D human pose estimation (HPE) currently faces several key challenges. Information from different viewpoints exhibits high variability due to complex environmental factors, posing difficulties in cross-view feature extraction and fusion. Additionally, multi-view 3D labeled pose data is rather scarce, and the impact of input 2D poses on 3D HPE accuracy has received little attention. To address these issues, we propose an Error-aware Self-supervised transformer framework for Multi-view 3D HPE (ESMformer). Firstly, we introduce a single-view multi-level feature extraction module to enhance pose features in individual viewpoints, which incorporates a novel relative attention mechanism for representative feature extraction at different levels. Subsequently, we develop multi-view intra-level and cross-level fusion modules to exploit spatio-temporal feature dependencies among human joints, and progressively fuse pose information from all views and levels. Furthermore, we explore an error-aware self-supervised learning strategy to reduce the model’s reliance on 3D pose annotations and mitigate the impact of incorrect 2D poses. This strategy adaptively selects reliable input 2D poses based on 3D pose prediction errors. Experiments on three popular benchmarks show that ESMformer achieves state-of-the-art results and maintains cost-effective computational complexity. Notably, ESMformer does not rely on any 3D pose annotations or prior human body knowledge, making it highly versatile and adaptable in practical applications.11The code and models are available at https://github.com/z0911k/ESMformer.

The Implication of Fragment-centric Mapping Strategy to Explore the First Selective Inhibitor Against Target Protein

Introduction: Identification of the first selective inhibitor, also called “hit molecules, " is crucial for developing drugs against a protein target. However, the crystal structures of protein-ligand complexes are usually not resolved in time due to the process's time-consuming and costly nature. However, it does not prevent scientists from understanding the binding modes’ urgent advantages and drawbacks of protein-ligand interaction to guide the optimization of hit molecules. Methods: Here, we have developed a pocket-centric computational strategy to facilitate a comprehensive understanding of the hit molecules against the protein target. Results: The results show that the pocket-centered mapping method not only allows for accurate prediction of the native docking pose and in-depth analysis of the binding mode but also has the potential of rapidly identifying partially unoccupied, unutilized, but targetable pockets to afford optimized hit molecules. We tested the strategy on the first selective inhibitor, epigallocatechin gallate (EGCG), against human arylacetamide deacetylase (AADAC). Molecular dynamics simulation and MM/PBSA binding energy calculation are used to verify the efficacy of the strategy. Conclusion: The results show that the pocket-centered mapping method not only allows for accurate prediction of the native docking pose and in-depth analysis of the binding mode but also has the potential of rapidly identifying partially unoccupied, unutilized, but targetable pockets to afford optimized hit molecules.

Human pose estimation in complex background videos via Transformer-based multi-scale feature integration

Human posture estimation is still a hot research topic. Previous algorithms based on traditional machine learning have difficulties in feature extraction and low fusion efficiency. To address these problems, we proposed a Transformer-based method. We combined three techniques, namely the Transformer-based feature extraction module, the multi-scale feature fusion module, and the occlusion processing mechanism, to capture the human pose. The Transformer-based feature extraction module uses the self-attention mechanism to extract key features from the input sequence, the multi-scale feature fusion module fuses feature information of different scales to enhance the perception ability of the model, and the occlusion processing mechanism can effectively handle occlusion in the data and effectively remove background interference. Our method has shown excellent performance through verification on the standard dataset Human3.6M and the wild video dataset, achieving accurate pose prediction in both complex actions and challenging samples.

PheSA: An Open-Source Tool for Pharmacophore-Enhanced Shape Alignment.

PheSA is an open-source pharmacophore- and shape-based screening and molecular alignment tool that is fully open-source as part of OpenChemLib. Supporting standard ligand-based screening, flexible refinement of alignments, and receptor-guided shape docking, PheSA is a very flexible tool and can be used for different use cases in structure-based drug design. We present the algorithm and different benchmark studies that investigate the screening performance and also the quality of the generated alignments and the pose prediction performance of the receptor-guided PheSA algorithm. An important finding is the effect of the type of similarity metric used for measuring screening enrichment (symmetric Tanimoto versus asymmetric Tversky), whereby we could observe improved enrichment rates by using Tversky. PheSA exhibits enrichments on the DUD-E that are on par with commercial methods.

Dual-COPE: A novel prior-based category-level object pose estimation network with dual Sim2Real unsupervised domain adaptation module

Category-level pose estimation offers the generalization ability to novel objects unseen during training, which has attracted increasing attention in recent years. Despite the advantage, annotating real-world data with pose label is intricate and laborious. Although using synthetic data with free annotations can greatly reduce training costs, the Synthetic-to-Real (Sim2Real) domain gap could result in a sharp performance decline on real-world test. In this paper, we propose Dual-COPE, a novel prior-based category-level object pose estimation method with dual Sim2Real domain adaptation to avoid expensive real pose annotations. First, we propose an estimation network featured with conjoined prior deformation and transformer-based matching to realize high-precision pose prediction. Upon that, an efficient dual Sim2Real domain adaptation module is further designed to reduce the feature distribution discrepancy between synthetic and real-world data both semantically and geometrically, thus maintaining superior performance on real-world test. Moreover, the adaptation module is loosely coupled with estimation network, allowing for easy integration with other methods without any additional inference overhead. Comprehensive experiments show that Dual-COPE outperforms existing unsupervised methods and achieves state-of-the-art precision under supervised settings.

Temporally enhanced graph convolutional network for hand tracking from an egocentric camera

We propose a robust 3D hand tracking system in various hand action environments, including hand-object interaction, which utilizes a single color image and a previous pose prediction as input. We observe that existing methods deterministically exploit temporal information in motion space, failing to address realistic diverse hand motions. Also, prior methods paid less attention to efficiency as well as robust performance, i.e., the balance issues between time and accuracy. The Temporally Enhanced Graph Convolutional Network (TE-GCN) utilizes a 2-stage framework to encode temporal information adaptively. The system establishes balance by adopting an adaptive GCN, which effectively learns the spatial dependency between hand mesh vertices. Furthermore, the system leverages the previous prediction by estimating the relevance across image features through the attention mechanism. The proposed method achieves state-of-the-art balanced performance on challenging benchmarks and demonstrates robust results on various hand motions in real scenes. Moreover, the hand tracking system is integrated into a recent HMD with an off-loading framework, achieving a real-time framerate while maintaining high performance. Our study improves the usability of a high-performance hand-tracking method, which can be generalized to other algorithms and contributes to the usage of HMD in everyday life. Our code with the HMD project will be available at https://github.com/UVR-WJCHO/TEGCN_on_Hololens2.

HydraScreen: A Generalizable Structure-Based Deep Learning Approach to Drug Discovery.

We propose HydraScreen, a deep-learning framework for safe and robust accelerated drug discovery. HydraScreen utilizes a state-of-the-art 3D convolutional neural network designed for the effective representation of molecular structures and interactions in protein-ligand binding. We designed an end-to-end pipeline for high-throughput screening and lead optimization, targeting applications in structure-based drug design. We assessed our approach using established public benchmarks based on the CASF-2016 core set, achieving top-tier results in affinity and pose prediction (Pearson's r = 0.86, RMSE = 1.15, Top-1 = 0.95). We introduced a novel approach for interaction profiling, aimed at detecting potential biases within both the model and data sets. This approach not only enhanced interpretability but also reinforced the impartiality of our methodology. Finally, we demonstrated HydraScreen's ability to generalize effectively across novel proteins and ligands through a temporal split. We also provide insights into potential avenues for future development aimed at enhancing the robustness of machine learning scoring functions. HydraScreen (accessible at http://hydrascreen.ro5.ai/paper) provides a user-friendly GUI and a public API, facilitating the easy-access assessment of protein-ligand complexes.

Quantum Mechanics-Based Fast and Reliable Prediction of Binding Pose Structures.

Predicting the binding poses of docking with an accurate estimation of binding energies is highly important but very challenging in computational drug design. A quantum mechanics (QM) calculation-based docking approach considering multiple conformations and orientations of the ligand is introduced here to tackle the problem. This QM docking consists of three steps: generating an ensemble of binding poses with a conventional docking simulation, computing the binding energies with self-consistent charge density functional theory tightly binding with dispersion correction (DFTB-D) to selecting the 10 top binding modes, and optimizing the selected binding mode structures using the ONIOM(DFTB:PM7) technique to determine the binding poses. The ONIOM(DFTB-D:PM6) docking approach is tested on 121 ligand-receptor biocomplexes with the crystal structures obtained from the Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB). The result shows that the new method is highly satisfactory for the accurate prediction of the binding poses. The new docking method should be beneficial to structure-based drug design.

Lightning Pose: improved animal pose estimation via semi-supervised learning, Bayesian ensembling and cloud-native open-source tools.

Contemporary pose estimation methods enable precise measurements of behavior via supervised deep learning with hand-labeled video frames. Although effective in many cases, the supervised approach requires extensive labeling and often produces outputs that are unreliable for downstream analyses. Here, we introduce 'Lightning Pose', an efficient pose estimation package with three algorithmic contributions. First, in addition to training on a few labeled video frames, we use many unlabeled videos and penalize the network whenever its predictions violate motion continuity, multiple-view geometry and posture plausibility (semi-supervised learning). Second, we introduce a network architecture that resolves occlusions by predicting pose on any given frame using surrounding unlabeled frames. Third, we refine the pose predictions post hoc by combining ensembling and Kalman smoothing. Together, these components render pose trajectories more accurate and scientifically usable. We released a cloud application that allows users to label data, train networks and process new videos directly from the browser.

G Protein-Coupled Receptor-Ligand Pose and Functional Class Prediction.

G protein-coupled receptor (GPCR) transmembrane protein family members play essential roles in physiology. Numerous pharmaceuticals target GPCRs, and many drug discovery programs utilize virtual screening (VS) against GPCR targets. Improvements in the accuracy of predicting new molecules that bind to and either activate or inhibit GPCR function would accelerate such drug discovery programs. This work addresses two significant research questions. First, do ligand interaction fingerprints provide a substantial advantage over automated methods of binding site selection for classical docking? Second, can the functional status of prospective screening candidates be predicted from ligand interaction fingerprints using a random forest classifier? Ligand interaction fingerprints were found to offer modest advantages in sampling accurate poses, but no substantial advantage in the final set of top-ranked poses after scoring, and, thus, were not used in the generation of the ligand-receptor complexes used to train and test the random forest classifier. A binary classifier which treated agonists, antagonists, and inverse agonists as active and all other ligands as inactive proved highly effective in ligand function prediction in an external test set of GPR31 and TAAR2 candidate ligands with a hit rate of 82.6% actual actives within the set of predicted actives.

Computational evaluation and benchmark study of 342 crystallographic holo-structures of SARS-CoV-2 Mpro enzyme

The coronavirus disease 19 pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has led to a global health crisis with millions of confirmed cases and related deaths. The main protease (Mpro) of SARS-CoV-2 is crucial for viral replication and presents an attractive target for drug development. Despite the approval of some drugs, the search for effective treatments continues. In this study, we systematically evaluated 342 holo-crystal structures of Mpro to identify optimal conformations for structure-based virtual screening (SBVS). Our analysis revealed limited structural flexibility among the structures. Three docking programs, AutoDock Vina, rDock, and Glide were employed to assess the efficiency of virtual screening, revealing diverse performances across selected Mpro structures. We found that the structures 5RHE, 7DDC, and 7DPU (PDB Ids) consistently displayed the lowest EF, AUC, and BEDROCK scores. Furthermore, these structures demonstrated the worst pose prediction results in all docking programs. Two structural differences contribute to variations in docking performance: the absence of the S1 subsite in 7DDC and 7DPU, and the presence of a subpocket in the S2 subsite of 7DDC, 7DPU, and 5RHE. These findings underscore the importance of selecting appropriate Mpro conformations for SBVS, providing valuable insights for advancing drug discovery efforts.