Pose Prediction Research Articles

Predicting bioactive conformations and binding modes of macrocycles

Macrocyclic compounds experience increasing interest in drug discovery. It is often thought that these large and chemically complex molecules provide promising candidates to address difficult targets and interfere with protein-protein interactions. From a computational viewpoint, these molecules are difficult to treat. For example, flexible docking of macrocyclic compounds is hindered by the limited ability of current docking approaches to optimize conformations of extended ring systems for pose prediction. Herein, we report predictions of bioactive conformations of macrocycles using conformational search and binding modes using docking. Conformational ensembles generated using specialized search technique of about 70% of the tested macrocycles contained accurate bioactive conformations. However, these conformations were difficult to identify on the basis of conformational energies. Moreover, docking calculations with limited ligand flexibility starting from individual low energy conformations rarely yielded highly accurate binding modes. In about 40% of the test cases, binding modes were approximated with reasonable accuracy. However, when conformational ensembles were subjected to rigid body docking, an increase in meaningful binding mode predictions to more than 50% of the test cases was observed. Electrostatic effects did not contribute to these predictions in a positive or negative manner. Rather, achieving shape complementarity at macrocycle-target interfaces was a decisive factor. In summary, a combined computational protocol using pre-computed conformational ensembles of macrocycles as a starting point for docking shows promise in modeling binding modes of macrocyclic compounds.

Improved pose and affinity predictions using different protocols tailored on the basis of data availability.

The D3R 2015 grand drug design challenge provided a set of blinded challenges for evaluating the applicability of our protocols for pose and affinity prediction. In the present study, we report the application of two different strategies for the two D3R protein targets HSP90 and MAP4K4. HSP90 is a well-studied target system with numerous co-crystal structures and SAR data. Furthermore the D3R HSP90 test compounds showed high structural similarity to existing HSP90 inhibitors in BindingDB. Thus, we adopted an integrated docking and scoring approach involving acombination of both pharmacophoric and heavy atom similarity alignments, local minimization and quantitative structure activity relationships modeling, resulting in the reasonable prediction of pose [with the root mean square deviation (RMSD) values of 1.75Å for mean pose 1, 1.417Å for the mean best pose and 1.85Å for the mean all poses] and affinity (ROC AUC=0.702 at 7.5 pIC50 cut-off and R=0.45 for 180 compounds). The second protein, MAP4K4, represents a novel system with limited SAR and co-crystal structure data and little structural similarity of the D3R MAP4K4 test compounds to known MAP4K4ligands. For this system, we implemented an exhaustive pose and affinity prediction protocol involving docking and scoring using the PLANTS software which considers side chain flexibility together with protein-ligand fingerprints analysis assisting in pose prioritization. This protocol through fares poorly in pose prediction (with the RMSD values of 4.346Å for mean pose 1, 4.69Å for mean best pose and 4.75Å for mean all poses) and produced reasonable affinity prediction (AUC=0.728 at 7.5 pIC50 cut-off and R=0.67 for 18 compounds, ranked 1st among 80 submissions).

D3R grand challenge 2015: Evaluation of protein-ligand pose and affinity predictions.

The Drug Design Data Resource (D3R) ran Grand Challenge 2015 between September 2015 and February 2016. Two targets served as the framework to test community docking and scoring methods: (1) HSP90, donated by AbbVie and the Community Structure Activity Resource (CSAR), and (2) MAP4K4, donated by Genentech. The challenges for both target datasets were conducted in two stages, with the first stage testing pose predictions and the capacity to rank compounds by affinity with minimal structural data; and the second stage testing methods for ranking compounds with knowledge of at least a subset of the ligand-protein poses. An additional sub-challenge provided small groups of chemically similar HSP90 compounds amenable to alchemical calculations of relative binding free energy. Unlike previous blinded Challenges, we did not provide cognate receptors or receptors prepared with hydrogens and likewise did not require a specified crystal structure to be used for pose or affinity prediction in Stage 1. Given the freedom to select from over 200 crystal structures of HSP90 in the PDB, participants employed workflows that tested not only core docking and scoring technologies, but also methods for addressing water-mediated ligand-protein interactions, binding pocket flexibility, and the optimal selection of protein structures for use in docking calculations. Nearly 40 participating groups submitted over 350 prediction sets for Grand Challenge 2015. This overview describes the datasets and the organization of the challenge components, summarizes the results across all submitted predictions, and considers broad conclusions that may be drawn from this collaborative community endeavor.

Predicting binding poses and affinities for protein - ligand complexes in the 2015 D3R Grand Challenge using a physical model with a statistical parameter estimation.

The 2015 D3R Grand Challenge provided an opportunity to test our new model for the binding free energy of small molecules, as well as to assess our protocol to predict binding poses for protein-ligand complexes. Our pose predictions were ranked 3-9 for the HSP90 dataset, depending on the assessment metric. For the MAP4K dataset the ranks are very dispersed and equal to 2-35, depending on the assessment metric, which does not provide any insight into the accuracy of the method. The main success of our pose prediction protocol was the re-scoring stage using the recently developed Convex-PL potential. We make a thorough analysis of our docking predictions made with AutoDock Vina and discuss the effect of the choice of rigid receptor templates, the number of flexible residues in the binding pocket, the binding pocket size, and the benefits of re-scoring. However, the main challenge was to predict experimentally determined binding affinities for two blind test sets. Our affinity prediction model consisted of two terms, a pairwise-additive enthalpy, and a non pairwise-additive entropy. We trained the free parameters of the model with a regularized regression using affinity and structural data from the PDBBind database. Our model performed very well on the training set, however, failed on the two test sets. We explain the drawback and pitfalls of our model, in particular in terms of relative coverage of the test set by the training set and missed dynamical properties from crystal structures, and discuss different routes to improve it.

DockBench as docking selector tool: the lesson learned from D3R Grand Challenge 2015.

Structure-based drug design (SBDD) has matured within the last two decades as a valuable tool for the optimization of low molecular weight lead compounds to highly potent drugs. The key step in SBDD requires knowledge of the three-dimensional structure of the target-ligand complex, which is usually determined by X-ray crystallography. In the absence of structural information for the complex, SBDD relies on the generation of plausible molecular docking models. However, molecular docking protocols suffer from inaccuracies in the description of the interaction energies between the ligand and the target molecule, and often fail in the prediction of the correct binding mode. In this context, the appropriate selection of the most accurate docking protocol is absolutely relevant for the final molecular docking result, even if addressing this point is absolutely not a trivial task. D3R Grand Challenge 2015 has represented a precious opportunity to test the performance of DockBench, an integrate informatics platform to automatically compare RMDS-based molecular docking performances of different docking/scoring methods. The overall performance resulted in the blind prediction are encouraging in particular for the pose prediction task, in which several complex were predicted with a sufficient accuracy for medicinal chemistry purposes.

A D3R prospective evaluation of machine learning for protein-ligand scoring.

We assess the performance of several machine learning-based scoring methods at protein-ligand pose prediction, virtual screening, and binding affinity prediction. The methods and the manner in which they were trained make them sufficiently diverse to evaluate the utility of various strategies for training set curation and binding pose generation, but they share a novel approach to classification in the context of protein-ligand scoring. Rather than explicitly using structural data such as affinity values or information extracted from crystal binding poses for training, we instead exploit the abundance of data available from high-throughput screening to approach the problem as one of discriminating binders from non-binders. We evaluate the performance of our various scoring methods in the 2015 D3R Grand Challenge and find that although the merits of some features of our approach remain inconclusive, our scoring methods performed comparably to a state-of-the-art scoring function that was fit to binding affinity data.

Optimal strategies for virtual screening of induced-fit and flexible target in the 2015 D3R Grand Challenge.

Induced fit or protein flexibility can make a given structure less useful for docking and/or scoring. The 2015 Drug Design Data Resource (D3R) Grand Challenge provided a unique opportunity to prospectively test optimal strategies for virtual screening in these type of targets: heat shock protein 90 (HSP90), a protein with multiple ligand-induced binding modes; and mitogen-activated protein kinase kinase kinase kinase 4 (MAP4K4), a kinase with a large flexible pocket. Using previously known co-crystal structures, we tested predictions from methods that keep the receptor structure fixed and used (a) multiple receptor/ligand co-crystals as binding templates for minimization or docking ("close"), (b) methods that align or dock to a single receptor ("cross"), and (c) a hybrid approach that chose from multiple bound ligands as initial templates for minimization to a single receptor ("min-cross"). Pose prediction using our "close" models resulted in average ligand RMSDs of 0.32 and 1.6Å for HSP90 and MAP4K4, respectively, the most accurate models of the community-wide challenge. On the other hand, affinity ranking using our "cross" methods performed well overall despite the fact that a fixed receptor cannot model ligand-induced structural changes,. In addition, "close" methods that leverage the co-crystals of the different binding modes of HSP90 also predicted the best affinity ranking. Our studies suggest that analysis of changes on the receptor structure upon ligand binding can help select an optimal virtual screening strategy.

Prospective evaluation of shape similarity based pose prediction method in D3R Grand Challenge 2015.

Evaluation of ligand three-dimensional (3D) shape similarity is one of the commonly used approaches to identify ligands similar to one or more known active compounds from a library of small molecules. Apart from using ligand shape similarity as a virtual screening tool, its role in pose prediction and pose scoring has also been reported. We have recently developed a method that utilizes ligand 3D shape similarity with known crystallographic ligands to predict binding poses of query ligands. Here, we report the prospective evaluation of our pose prediction method through the participation in drug design data resource (D3R) Grand Challenge 2015. Our pose prediction method was used to predict binding poses of heat shock protein 90 (HSP90) and mitogen activated protein kinase kinase kinase kinase (MAP4K4) ligands and it was able to predict the pose within 2Å root mean square deviation (RMSD) either as the top pose or among the best of five poses in a majority of cases. Specifically for HSP90 protein, a median RMSD of 0.73 and 0.68Å was obtained for the top and the best of five predictions respectively. For MAP4K4 target, although the median RMSD for our top prediction was only 2.87Å but the median RMSD of 1.67Å for the best of five predictions was well within the limit for successful prediction. Furthermore, the performance of our pose prediction method for HSP90 and MAP4K4 ligands was always among the top five groups. Particularly, for MAP4K4 protein our pose prediction method was ranked number one both in terms of mean and median RMSD when the best of five predictions were considered. Overall, our D3R Grand Challenge 2015 results demonstrated that ligand 3D shape similarity with the crystal ligand is sufficient to predict binding poses of new ligands with acceptable accuracy.

Blinded predictions of binding modes and energies of HSP90-α ligands for the 2015 D3R grand challenge

In the framework of the 2015 D3R inaugural grand challenge, blind binding pose and affinity predictions were performed for a set of 180 ligands of the Heat Shock Protein HSP90-α protein, a relevant cancer target. Spectral clustering was used to rapidly identify alternative binding site conformations in publicly available crystallographic HSP90-α structures. Subsequently, multiple docking and scoring protocols employing the software Autodock Vina and rDock were applied to predict binding modes and rank order ligands. Alchemical free energy calculations were performed with the software FESetup and Sire/OpenMM to predict binding affinities for three congeneric series subsets. Some of the protocols used here were ranked among the top submissions according to most of the evaluation metrics. Docking performance was excellent, but the scoring results were disappointing. A critical assessment of the results is reported, as well as suggestions for future similar competitions.

Protein-Ligand Informatics Force Field (PLIff): Toward a Fully Knowledge Driven "Force Field" for Biomolecular Interactions.

The Protein Data Bank (PDB) contains a wealth of data on nonbonded biomolecular interactions. If this information could be distilled down to nonbonded interaction potentials, these would have some key advantages over standard force fields. However, there are some important outstanding issues to address in order to do this successfully. This paper introduces the protein-ligand informatics "force field", PLIff, which begins to address these key challenges ( https://bitbucket.org/AstexUK/pli ). As a result of their knowledge-based nature, the next-generation nonbonded potentials that make up PLIff automatically capture a wide range of interaction types, including special interactions that are often poorly described by standard force fields. We illustrate how PLIff may be used in structure-based design applications, including interaction fields, fragment mapping, and protein-ligand docking. PLIff performs at least as well as state-of-the art scoring functions in terms of pose predictions and ranking compounds in a virtual screening context.

A pose prediction approach based on ligand 3D shape similarity

Molecular docking predicts the best pose of a ligand in the target protein binding site by sampling and scoring numerous conformations and orientations of the ligand. Failures in pose prediction are often due to either insufficient sampling or scoring function errors. To improve the accuracy of pose prediction by tackling the sampling problem, we have developed a method of pose prediction using shape similarity. It first places a ligand conformation of the highest 3D shape similarity with known crystal structure ligands into protein binding site and then refines the pose by repacking the side-chains and performing energy minimization with a Monte Carlo algorithm. We have assessed our method utilizing CSARdock 2012 and 2014 benchmark exercise datasets consisting of co-crystal structures from eight proteins. Our results revealed that ligand 3D shape similarity could substitute conformational and orientational sampling if at least one suitable co-crystal structure is available. Our method identified poses within 2 Å RMSD as the top-ranking pose for 85.7% of the test cases. The median RMSD for our pose prediction method was found to be 0.81 Å and was better than methods performing extensive conformational and orientational sampling within target protein binding sites. Furthermore, our method was better than similar methods utilizing ligand 3D shape similarity for pose prediction.

ON APPROACHING HEURISTIC WEIGHT MASK TO ENHANCE LBP-BASED PROFILE FACE RECOGNITION SYSTEM

In this paper, a LBP- based profile face recognizer for identifying a query face image is proposed. A robust heuristic approach is proposed to improve the weight set of LBP face recognizers. Our method can be combined with other variants of LBP feature extraction to improve their accuracy, because it can help to avoid the local maxima. For identifying a profile face image, an appropriate framework predicting the pose range of image to choose the suitable gallery for recognition is presented. It helps us to avoid error propagation when pose prediction returns false results. Recognizing image with gallery set of surrounding pose range increases chance of input image to match its subject’s sample images. Experimental results on the FERET database show the effectiveness of these methods.

Machine learning optimization of cross docking accuracy

Performance of small molecule automated docking programs has conceptually been divided into docking -, scoring -, ranking - and screening power, which focuses on the crystal pose prediction, affinity prediction, ligand ranking and database screening capabilities of the docking program, respectively. Benchmarks show that different docking programs can excel in individual benchmarks which suggests that the scoring function employed by the programs can be optimized for a particular task. Here the scoring function of Smina is re-optimized towards enhancing the docking power using a supervised machine learning approach and a manually curated database of ligands and cross docking receptor pairs. The optimization method does not need associated binding data for the receptor-ligand examples used in the data set and works with small train sets. The re-optimization of the weights for the scoring function results in a similar docking performance with regard to docking power towards a cross docking test set. A ligand decoy based benchmark indicates a better discrimination between poses with high and low RMSD. The reported parameters for Smina are compatible with Autodock Vina and represent ready-to-use alternative parameters for researchers who aim at pose prediction rather than affinity prediction.

Fully automatic person segmentation in unconstrained video using spatio-temporal conditional random fields

The segmentation of objects and people in particular is an important problem in computer vision. In this paper, we focus on automatically segmenting a person from challenging video sequences in which we place no constraint on camera viewpoint, camera motion or the movements of a person in the scene. Our approach uses the most confident predictions from a pose detector as a form of anchor or keyframe stick figure prediction which helps guide the segmentation of other more challenging frames in the video. Since even state of the art pose detectors are unreliable on many frames –especially given that we are interested in segmentations with no camera or motion constraints –only the poses or stick figure predictions for frames with the highest confidence in a localized temporal region anchor further processing. The stick figure predictions within confident keyframes are used to extract color, position and optical flow features. Multiple conditional random fields (CRFs) are used to process blocks of video in batches, using a two dimensional CRF for detailed keyframe segmentation as well as 3D CRFs for propagating segmentations to the entire sequence of frames belonging to batches. Location information derived from the pose is also used to refine the results. Importantly, no hand labeled training data is required by our method. We discuss the use of a continuity method that reuses learnt parameters between batches of frames and show how pose predictions can also be improved by our model. We provide an extensive evaluation of our approach, comparing it with a variety of alternative grab cut based methods and a prior state of the art method. We also release our evaluation data to the community to facilitate further experiments. We find that our approach yields state of the art qualitative and quantitative performance compared to prior work and more heuristic alternative approaches.

WScore: A Flexible and Accurate Treatment of Explicit Water Molecules in Ligand-Receptor Docking.

We have developed a new methodology for protein-ligand docking and scoring, WScore, incorporating a flexible description of explicit water molecules. The locations and thermodynamics of the waters are derived from a WaterMap molecular dynamics simulation. The water structure is employed to provide an atomic level description of ligand and protein desolvation. WScore also contains a detailed model for localized ligand and protein strain energy and integrates an MM-GBSA scoring component with these terms to assess delocalized strain of the complex. Ensemble docking is used to take into account induced fit effects on the receptor conformation, and protein reorganization free energies are assigned via fitting to experimental data. The performance of the method is evaluated for pose prediction, rank ordering of self-docked complexes, and enrichment in virtual screening, using a large data set of PDB complexes and compared with the Glide SP and Glide XP models; significant improvements are obtained.

Pictorial Human Spaces: A Computational Study on the Human Perception of 3D Articulated Poses

Human motion analysis in images and video, with its deeply inter-related 2D and 3D inference components, is a central computer vision problem. Yet, there are no studies that reveal how humans perceive other people in images and how accurate they are. In this paper we aim to unveil some of the processing--as well as the levels of accuracy--involved in the 3D perception of people from images by assessing the human performance. Moreover, we reveal the quantitative and qualitative differences between human and computer performance when presented with the same visual stimuli and show that metrics incorporating human perception can produce more meaningful results when integrated into automatic pose prediction algorithms. Our contributions are: (1) the construction of an experimental apparatus that relates perception and measurement, in particular the visual and kinematic performance with respect to 3D ground truth when the human subject is presented an image of a person in a given pose; (2) the creation of a dataset containing images, articulated 2D and 3D pose ground truth, as well as synchronized eye movement recordings of human subjects, shown a variety of human body configurations, both easy and difficult, as well as their `re-enacted' 3D poses; (3) quantitative analysis revealing the human performance in 3D pose re-enactment tasks, the degree of stability in the visual fixation patterns of human subjects, and the way it correlates with different poses; (4) extensive analysis on the differences between human re-enactments and poses produced by an automatic system when presented with the same visual stimuli; (5) an approach to learning perceptual metrics that, when integrated into visual sensing systems, produces more stable and meaningful results.

Exploiting View-Specific Appearance Similarities Across Classes for Zero-Shot Pose Prediction: A Metric Learning Approach

Viewpoint estimation, especially in case of multiple object classes, remains an important and challenging problem. First, objects under different views undergo extreme appearance variations, often making within-class variance larger than between-class variance. Second, obtaining precise ground truth for real-world images, necessary for training supervised viewpoint estimation models, is extremely difficult and time consuming. As a result, annotated data is often available only for a limited number of classes. Hence it is desirable to share viewpoint information across classes. Additional complexity arises from unaligned pose labels between classes, i.e. a side view of a car might look more like a frontal view of a toaster, than its side view. To address these problems, we propose a metric learning approach for joint class prediction and pose estimation. Our approach allows to circumvent the problem of viewpoint alignment across multiple classes, and does not require dense viewpoint labels. Moreover, we show, that the learned metric generalizes to new classes, for which the pose labels are not available, and therefore makes it possible to use only partially annotated training sets, relying on the intrinsic similarities in the viewpoint manifolds. We evaluate our approach on two challenging multi-class datasets, 3DObjects and PASCAL3D+.

Sticky Projections-A Model-Based Approach to Interactive Shader Lamps Tracking.

Shader lamps can augment physical objects with projected virtual replications using a camera-projector system, provided that the physical and virtual object are well registered to each other. Precise registration and tracking has been a cumbersome and intrusive process in the past. In this paper, we present a new method for tracking complex-shaped physical objects interactively. In contrast to previous approaches our system is mobile and makes solely use of the projection of the virtual replication to track the physical object and "stick" the projection to it. Our method consists of two stages, a fast pose initialization based on structured light patterns and a non-intrusive frame-by-frame tracking based on features detected in the projection. During the tracking phase, a radiometrically corrected virtual camera view based on the current pose prediction is rendered and compared to the captured image. Matched features are triangulated providing a sparse set of surface points that is robustly aligned to the virtual model. The alignment transformation serves as an input for the new pose prediction. Detailed experiments including the evaluation of the overlay accuracy show that our approach can accurately and robustly track complex objects at interactive rates.

Tanker pose prediction based on modified unscented Kalman filter

Tanker pose prediction based on modified unscented Kalman filter

Single-Point Mutation with a Rotamer Library Toolkit: Toward Protein Engineering.

Protein engineers have long been hard at work to harness biocatalysts as a natural source of regio-, stereo-, and chemoselectivity in order to carry out chemistry (reactions and/or substrates) not previously achieved with these enzymes. The extreme labor demands and exponential number of mutation combinations have induced computational advances in this domain. The first step in our virtual approach is to predict the correct conformations upon mutation of residues (i.e., rebuilding side chains). For this purpose, we opted for a combination of molecular mechanics and statistical data. In this work, we have developed automated computational tools to extract protein structural information and created conformational libraries for each amino acid dependent on a variable number of parameters (e.g., resolution, flexibility, secondary structure). We have also developed the necessary tool to apply the mutation and optimize the conformation accordingly. For side-chain conformation prediction, we obtained overall average root-mean-square deviations (RMSDs) of 0.91 and 1.01 Å for the 18 flexible natural amino acids within two distinct sets of over 3000 and 1500 side-chain residues, respectively. The commonly used dihedral angle differences were also evaluated and performed worse than the state of the art. These two metrics are also compared. Furthermore, we generated a family-specific library for kinases that produced an average 2% lower RMSD upon side-chain reconstruction and a residue-specific library that yielded a 17% improvement. Ultimately, since our protein engineering outlook involves using our docking software, Fitted/Impacts, we applied our mutation protocol to a benchmarked data set for self- and cross-docking. Our side-chain reconstruction does not hinder our docking software, demonstrating differences in pose prediction accuracy of approximately 2% (RMSD cutoff metric) for a set of over 200 protein/ligand structures. Similarly, when docking to a set of over 100 kinases, side-chain reconstruction (using both general and biased conformation libraries) had minimal detriment to the docking accuracy.