Conformer Generation Methods Research Articles

Deep contrastive learning of molecular conformation for efficient property prediction.

Data-driven deep learning algorithms provide accurate prediction of high-level quantum-chemical molecular properties. However, their inputs must be constrained to the same quantum-chemical level of geometric relaxation as the training dataset, limiting their flexibility. Adopting alternative cost-effective conformation generative methods introduces domain-shift problems, deteriorating prediction accuracy. Here we propose a deep contrastive learning-based domain-adaptation method called Local Atomic environment Contrastive Learning (LACL). LACL learns to alleviate the disparities in distribution between the two geometric conformations by comparing different conformation-generation methods. We found that LACL forms a domain-agnostic latent space that encapsulates the semantics of an atom's local atomic environment. LACL achieves quantum-chemical accuracy while circumventing the geometric relaxation bottleneck and could enable future application scenarios such as inverse molecular engineering and large-scale screening. Our approach is also generalizable from small organic molecules to long chains of biological and pharmacological molecules.

SDEGen: learning to evolve molecular conformations from thermodynamic noise for conformation generation.

Generation of representative conformations for small molecules is a fundamental task in cheminformatics and computer-aided drug discovery, but capturing the complex distribution of conformations that contains multiple low energy minima is still a great challenge. Deep generative modeling, aiming to learn complex data distributions, is a promising approach to tackle the conformation generation problem. Here, inspired by stochastic dynamics and recent advances in generative modeling, we developed SDEGen, a novel conformation generation model based on stochastic differential equations. Compared with existing conformation generation methods, it enjoys the following advantages: (1) high model capacity to capture multimodal conformation distribution, thereby searching for multiple low-energy conformations of a molecule quickly, (2) higher conformation generation efficiency, almost ten times faster than the state-of-the-art score-based model, ConfGF, and (3) a clear physical interpretation to learn how a molecule evolves in a stochastic dynamics system starting from noise and eventually relaxing to the conformation that falls in low energy minima. Extensive experiments demonstrate that SDEGen has surpassed existing methods in different tasks for conformation generation, interatomic distance distribution prediction, and thermodynamic property estimation, showing great potential for real-world applications.

Molecular Geometry Prediction using a Deep Generative Graph Neural Network

A molecule’s geometry, also known as conformation, is one of a molecule’s most important properties, determining the reactions it participates in, the bonds it forms, and the interactions it has with other molecules. Conventional conformation generation methods minimize hand-designed molecular force field energy functions that are often not well correlated with the true energy function of a molecule observed in nature. They generate geometrically diverse sets of conformations, some of which are very similar to the lowest-energy conformations and others of which are very different. In this paper, we propose a conditional deep generative graph neural network that learns an energy function by directly learning to generate molecular conformations that are energetically favorable and more likely to be observed experimentally in data-driven manner. On three large-scale datasets containing small molecules, we show that our method generates a set of conformations that on average is far more likely to be close to the corresponding reference conformations than are those obtained from conventional force field methods. Our method maintains geometrical diversity by generating conformations that are not too similar to each other, and is also computationally faster. We also show that our method can be used to provide initial coordinates for conventional force field methods. On one of the evaluated datasets we show that this combination allows us to combine the best of both methods, yielding generated conformations that are on average close to reference conformations with some very similar to reference conformations.

Bayesian optimization for conformer generation

Generating low-energy molecular conformers is a key task for many areas of computational chemistry, molecular modeling and cheminformatics. Most current conformer generation methods primarily focus on generating geometrically diverse conformers rather than finding the most probable or energetically lowest minima. Here, we present a new stochastic search method called the Bayesian optimization algorithm (BOA) for finding the lowest energy conformation of a given molecule. We compare BOA with uniform random search, and systematic search as implemented in Confab, to determine which method finds the lowest energy. Energetic difference, root-mean-square deviation, and torsion fingerprint deviation are used to quantify the performance of the conformer search algorithms. In general, we find BOA requires far fewer evaluations than systematic or uniform random search to find low-energy minima. For molecules with four or more rotatable bonds, Confab typically evaluates 10^{4} (median) conformers in its search, while BOA only requires 10^{2} energy evaluations to find top candidates. Despite using evaluating fewer conformers, 20–40% of the time BOA finds lower-energy conformations than a systematic Confab search for molecules with four or more rotatable bonds.

Shape similarity guided pose prediction: lessons from D3R Grand Challenge 3.

To extend the utility of ligand 3D shape similarity into pose prediction and virtual screening, we have previously developed CDVS and PoPSS methods. Both of them utilize ligand 3D shape similarity with the crystallographic ligands to improve pose prediction. While CDVS utilizes shape similarity to select suitable receptor structures for molecular docking, PoPSS places a ligand conformation of the highest shape similarity with crystal ligands into the target protein binding pocket which is then refined by side-chain repacking and Monte Carlo energy minimization. Analyses of PoPSS revealed some drawbacks in ligand conformation generation and the scoring scheme used. Moreover, as PoPSS does not sample the ligand conformation after placing it in the binding pocket, it relies solely on conformation generation methods to produce native like conformations. To address these limitations of PoPSS method, we report here a modified approach named as PoPSS-Lite, where side-chain repacking was replaced by a simple grid-based energy minimization. This modification also allowed the sampling of terminal functional groups while keeping the core scaffold fixed. Furthermore, shape similarity calculations were improved by increasing the number of ligand conformations and using a different similarity metric. The performance of PoPSS-Lite was prospectively evaluated in D3R GC3. Comparison of PoPSS-Lite demonstrated superior performance over PoPSS and CDVS with lower mean and median RMSDs. Furthermore, comparison with other D3R GC3 pose prediction submissions revealed top performance for PoPSS-Lite. Our D3R GC3 result extends our perspective that ligand 3D shape similarity with known crystallographic information can be successfully used to predict the binding pose of ligands with unknown binding modes. Our D3R GC3 results further highlight the necessity for improvement in conformer generation methods in order to improve shape similarity guided pose prediction.

Efficient conformational ensemble generation of protein-bound peptides

Conformation generation of protein-bound peptides is critical for the determination of protein–peptide complex structures. Despite significant progress in conformer generation of small molecules, few methods have been developed for modeling protein-bound peptide conformations. Here, we have developed a fast de novo peptide modeling algorithm, referred to as MODPEP, for conformational sampling of protein-bound peptides. Given a sequence, MODPEP builds the peptide 3D structure from scratch by assembling amino acids or helix fragments based on constructed rotamer and helix libraries. The MODPEP algorithm was tested on a diverse set of 910 experimentally determined protein-bound peptides with 3–30 amino acids from the PDB and obtained an average accuracy of 1.90 Å when 200 conformations were sampled for each peptide. On average, MODPEP obtained a success rate of 74.3% for all the 910 peptides and ≥ 90% for short peptides with 3–10 amino acids in reproducing experimental protein-bound structures. Comparative evaluations of MODPEP with three other conformer generation methods, PEP-FOLD3, RDKit, and Balloon, have also been performed in both accuracy and success rate. MODPEP is fast and can generate 100 conformations for less than one second. The fast MODPEP will be beneficial for large-scale de novo modeling and docking of peptides. The MODPEP program and libraries are available for download at http://huanglab.phys.hust.edu.cn/.

Examining the predictive accuracy of the novel 3D N-linear algebraic molecular codifications on benchmark datasets.

BackgroundRecently, novel 3D alignment-free molecular descriptors (also known as QuBiLS-MIDAS) based on two-linear, three-linear and four-linear algebraic forms have been introduced. These descriptors codify chemical information for relations between two, three and four atoms by using several (dis-)similarity metrics and multi-metrics. Several studies aimed at assessing the quality of these novel descriptors have been performed. However, a deeper analysis of their performance is necessary. Therefore, in the present manuscript an assessment and statistical validation of the performance of these novel descriptors in QSAR studies is performed.ResultsTo this end, eight molecular datasets (angiotensin converting enzyme, acetylcholinesterase inhibitors, benzodiazepine receptor, cyclooxygenase-2 inhibitors, dihydrofolate reductase inhibitors, glycogen phosphorylase b, thermolysin inhibitors, thrombin inhibitors) widely used as benchmarks in the evaluation of several procedures are utilized. Three to nine variable QSAR models based on Multiple Linear Regression are built for each chemical dataset according to the original division into training/test sets. Comparisons with respect to leave-one-out cross-validation correlation coefficientsleft( {Q_{loo}^{2} } right) reveal that the models based on QuBiLS-MIDAS indices possess superior predictive ability in 7 of the 8 datasets analyzed, outperforming methodologies based on similar or more complex techniques such as: Partial Least Square, Neural Networks, Support Vector Machine and others. On the other hand, superior external correlation coefficientsleft( {Q_{ext}^{2} } right) are attained in 6 of the 8 test sets considered, confirming the good predictive power of the obtained models. For the Q_{ext}^{2} values non-parametric statistic tests were performed, which demonstrated that the models based on QuBiLS-MIDAS indices have the best global performance and yield significantly better predictions in 11 of the 12 QSAR procedures used in the comparison. Lastly, a study concerning to the performance of the indices according to several conformer generation methods was performed. This demonstrated that the quality of predictions of the QSAR models based on QuBiLS-MIDAS indices depend on 3D structure generation method considered, although in this preliminary study the results achieved do not present significant statistical differences among them.ConclusionsAs conclusions it can be stated that the QuBiLS-MIDAS indices are suitable for extracting structural information of the molecules and thus, constitute a promissory alternative to build models that contribute to the prediction of pharmacokinetic, pharmacodynamics and toxicological properties on novel compounds.Graphical abstractComparative graphical representation of the performance of the novel QuBiLS-MIDAS 3D-MDs with respect to other methodologies in QSAR modeling of eight chemical datasetsElectronic supplementary materialThe online version of this article (doi:10.1186/s13321-016-0122-x) contains supplementary material, which is available to authorized users.

Freely Available Conformer Generation Methods: How Good Are They?

Conformer generation has important implications in cheminformatics, particularly in computational drug discovery where the quality of conformer generation software may affect the outcome of a virtual screening exercise. We examine the performance of four freely available small molecule conformer generation tools (Balloon, Confab, Frog2, and RDKit) alongside a commercial tool (MOE). The aim of this study is 3-fold: (i) to identify which tools most accurately reproduce experimentally determined structures; (ii) to examine the diversity of the generated conformational set; and (iii) to benchmark the computational time expended. These aspects were tested using a set of 708 drug-like molecules assembled from the OMEGA validation set and the Astex Diverse Set. These molecules have varying physicochemical properties and at least one known X-ray crystal structure. We found that RDKit and Confab are statistically better than other methods at generating low rmsd conformers to the known structure. RDKit is particularly suited for less flexible molecules while Confab, with its systematic approach, is able to generate conformers which are geometrically closer to the experimentally determined structure for molecules with a large number of rotatable bonds (≥10). In our tests RDKit also resulted as the second fastest method after Frog2. In order to enhance the performance of RDKit, we developed a postprocessing algorithm to build a diverse and representative set of conformers which also contains a close conformer to the known structure. Our analysis indicates that, with postprocessing, RDKit is a valid free alternative to commercial, proprietary software.

Cyndi: a multi-objective evolution algorithm based method for bioactive molecular conformational generation

BackgroundConformation generation is a ubiquitous problem in molecule modelling. Many applications require sampling the broad molecular conformational space or perceiving the bioactive conformers to ensure success. Numerous in silico methods have been proposed in an attempt to resolve the problem, ranging from deterministic to non-deterministic and systemic to stochastic ones. In this work, we described an efficient conformation sampling method named Cyndi, which is based on multi-objective evolution algorithm.ResultsThe conformational perturbation is subjected to evolutionary operation on the genome encoded with dihedral torsions. Various objectives are designated to render the generated Pareto optimal conformers to be energy-favoured as well as evenly scattered across the conformational space. An optional objective concerning the degree of molecular extension is added to achieve geometrically extended or compact conformations which have been observed to impact the molecular bioactivity (J Comput -Aided Mol Des 2002, 16: 105–112). Testing the performance of Cyndi against a test set consisting of 329 small molecules reveals an average minimum RMSD of 0.864 Å to corresponding bioactive conformations, indicating Cyndi is highly competitive against other conformation generation methods. Meanwhile, the high-speed performance (0.49 ± 0.18 seconds per molecule) renders Cyndi to be a practical toolkit for conformational database preparation and facilitates subsequent pharmacophore mapping or rigid docking. The copy of precompiled executable of Cyndi and the test set molecules in mol2 format are accessible in Additional file 1.ConclusionOn the basis of MOEA algorithm, we present a new, highly efficient conformation generation method, Cyndi, and report the results of validation and performance studies comparing with other four methods. The results reveal that Cyndi is capable of generating geometrically diverse conformers and outperforms other four multiple conformer generators in the case of reproducing the bioactive conformations against 329 structures. The speed advantage indicates Cyndi is a powerful alternative method for extensive conformational sampling and large-scale conformer database preparation.