Ultrafast shape recognition to search compound databases for similar molecular shapes

Development of a novel representation of drug 3D structures and enhancement of the TSR-based method for probing drug and target interactions

Understanding the mechanisms underlying interactions between drugs and target proteins is critical for drug discovery. In our earlier studies, we introduced the Triangular Spatial Relationship (TSR)-based algorithm, which enables the representation of a protein's 3D structure as a vector of integers (TSR keys). These TSR keys correspond to substructures of the 3D structure of a protein and are computed based on the triangles constructed by all possible triples of Cα atoms within the protein. In this study, we report on a new TSR-based algorithm for probing drug and target interactions. Specifically, we have extended the previous algorithm in three novel directions: TSR keys for representing the 3D structure of a drug or a ligand, cross TSR keys between drugs and their targets and intra-residual TSR keys for phosphorylated amino acids. The outcomes illustrate the key contributions as follows: (i) The TSR-based method, which uses the TSR keys as features, is unique in its capability to interpret hierarchical relationships of drugs as well as drug - target complexes using common and specific TSR keys. (ii) The method can distinguish not only the binding sites from the rest of the protein structures, but also the binding sites of primary targets from those of off-targets. (iii) The method has the potential to correlate the 3D structures of drugs with their functions. (iv) Representation of 3D structures by TSR keys has its unique advantage in terms of ease of making searching for similar substructures across structure datasets easier. In summary, this study presents a novel computational methodology, with significant advantages, for providing insights into the mechanism underlying drug and target interactions.

Collision-Induced Dissociation of Fucose and Identification of Anomericity

Structural determination of carbohydrates using mass spectrometry remains challenging, particularly, the differentiation of anomeric configurations. In this work, we studied the collision-induced dissociation (CID) mechanisms of sodiated α- and β-l-fucose using an experimental method and quantum chemistry calculations. The calculations show that α-l-fucose is more likely to undergo dehydration due to the fact that O1 and O2 are on the same side of the sugar ring. In contrast, β-l-fucose is more prone to the ring-opening reaction because more OH groups are on the same side of the sugar ring as O1. These differences suggest a higher preference for the dehydration reaction in sodiated α-l-fucose but a lower preference for ring-opening compared to that of β-l-fucose. The calculation results, which are used to assign the CID mass spectra of α- and β-l-fucose separated by high-performance liquid chromatography, are supported by the fucose produced from the CID of disaccharides Fuc-β-(1 → 3)-GlcNAc and Fuc-α-(1 → 4)-GlcNAc. This study demonstrates that the correlation of cis- and trans-configurations of O1 and O2 to the relative branching ratios of dehydration and cross-ring dissociation in CID, observed in aldohexose and ketohexose in the pyranose form, can be extended to deoxyhexoses for anomericity determination.

AutoDock-SS: AutoDock for Multiconformational Ligand-Based Virtual Screening

Ligand-based virtual screening (LBVS) can be pivotal for identifying potential drug leads, especially when the target protein's structure is unknown. However, current LBVS methods are limited in their ability to consider the ligand conformational flexibility. This study presents AutoDock-SS (Similarity Searching), which adapts protein-ligand docking for use in LBVS. AutoDock-SS integrates novel ligand-based grid maps and AutoDock-GPU into a novel three-dimensional LBVS workflow. Unlike other approaches based on pregenerated conformer libraries, AutoDock-SS's built-in conformational search optimizes conformations dynamically based on the reference ligand, thus providing a more accurate representation of relevant ligand conformations. AutoDock-SS supports two modes: single and multiple ligand queries, allowing for the seamless consideration of multiple reference ligands. When tested on the Directory of Useful Decoys─Enhanced (DUD-E) data set, AutoDock-SS surpassed alternative 3D LBVS methods, achieving a mean AUROC of 0.775 and an EF1% of 25.72 in single-reference mode. The multireference mode, evaluated on the augmented DUD-E+ data set, demonstrated superior accuracy with a mean AUROC of 0.843 and an EF1% of 34.59. This enhanced performance underscores AutoDock-SS's ability to treat compounds as conformationally flexible while considering the ligand's shape, pharmacophore, and electrostatic potential, expanding the potential of LBVS methods.

Advances in machine intelligence-driven virtual screening approaches for big-data

Virtual screening (VS) is an integral and ever-evolving domain of drug discovery framework. The VS is traditionally classified into ligand-based (LB) and structure-based (SB) approaches. Machine intelligence or artificial intelligence has wide applications in the drug discovery domain to reduce time and resource consumption. In combination with machine intelligence algorithms, VS has emerged into revolutionarily progressive technology that learns within robust decision orders for data curation and hit molecule screening from large VS libraries in minutes or hours. The exponential growth of chemical and biological data has evolved as "big-data" in the public domain demands modern and advanced machine intelligence-driven VS approaches to screen hit molecules from ultra-large VS libraries. VS has evolved from an individual approach (LB and SB) to integrated LB and SB techniques to explore various ligand and target protein aspects for the enhanced rate of appropriate hit molecule prediction. Current trends demand advanced and intelligent solutions to handle enormous data in drug discovery domain for screening and optimizing hits or lead with fewer or no false positive hits. Following the big-data drift and tremendous growth in computational architecture, we presented this review. Here, the article categorized and emphasized individual VS techniques, detailed literature presented for machine learning implementation, modern machine intelligence approaches, and limitations and deliberated the future prospects.

PL-PatchSurfer3: Improved Structure-Based Virtual Screening for Structure Variation Using 3D Zernike Descriptors

Structure-based virtual screening (SBVS) is a widely used method in silico drug discovery. It necessitates a receptor structure or binding site to predict the binding pose and fitness of a ligand. Therefore, the performance of the SBVS is affected by the protein conformation. The most frequently used method in SBVS is the protein-ligand docking program, which utilizes atomic distance-based scoring functions. Hence, they are highly prone to sensitivity towards variation in receptor structure, and it is reported that the conformational change significantly drops the performance of the docking program. To address the problem, we have introduced a novel program of SBVS, named PL-PatchSurfer. This program makes use of molecular surface patches and the Zernike descriptor. The surfaces of the pocket and ligand are segmented into several patches by the program. These patches are then mapped with physico-chemical properties such as shape and electrostatic potential before being converted into the Zernike descriptor, which is rotationally invariant. A complementarity between the protein and the ligand is assessed by comparing the descriptors and geometric distribution of the patches in the molecules. A benchmarking study showed that PL-PatchSurfer2 was able to screen active molecules regardless of the receptor structure change with fast speed. However, the program could not achieve high performance for the targets that the hydrogen bonding feature is important such as nuclear hormone receptors. In this paper, we present the newer version of PL-PatchSurfer, PL-PatchSurfer3, which incorporates two new features: a change in the definition of hydrogen bond complementarity and consideration of visibility that contains curvature information of a patch. Our evaluation demonstrates that the new program outperforms its predecessor and other SBVS methods while retaining its characteristic tolerance to receptor structure changes. Interested individuals can access the program at kiharalab.org/plps3.

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.

Repurposing Drugs for Inhibition against ALDH2 via a 2D/3D Ligand-Based Similarity Search and Molecular Simulation

Aldehyde dehydrogenase-2 (ALDH2) is a crucial enzyme participating in intracellular aldehyde metabolism and is acknowledged as a potential therapeutic target for the treatment of alcohol use disorder and other addictive behaviors. Using previously reported ALDH2 inhibitors of Daidzin, CVT-10216, and CHEMBL114083 as reference molecules, here we perform a ligand-based virtual screening of world-approved drugs via 2D/3D similarity search methods, followed by the assessments of molecular docking, toxicity prediction, molecular simulation, and the molecular mechanics Poisson-Boltzmann surface area (MM-PBSA) analysis. The 2D molecular fingerprinting of ECFP4 and FCFP4 and 3D molecule-shape-based USRCAT methods show good performances in selecting compounds with a strong binding behavior with ALDH2. Three compounds of Zeaxanthin (q = 0), Troglitazone (q = 0), and Sequinavir (q = +1 e) are singled out as potential inhibitors; Zeaxanthin can only be hit via USRCAT. These drugs displayed a stronger binding strength compared to the reported potent inhibitor CVT-10216. Sarizotan (q = +1 e) and Netarsudil (q = 0/+1 e) displayed a strong binding strength with ALDH2 as well, whereas they displayed a shallow penetration into the substrate-binding tunnel of ALDH2 and could not fully occupy it. This likely left a space for substrate binding, and thus they were not ideal inhibitors. The MM-PBSA results indicate that the selected negatively charged compounds from the similarity search and Vina scoring are thermodynamically unfavorable, mainly due to electrostatic repulsion with the receptor (q = -6 e for ALDH2). The electrostatic attraction with positively charged compounds, however, yielded very strong binding results with ALDH2. These findings reveal a deficiency in the modeling of electrostatic interactions (in particular, between charged moieties) in the virtual screening via the 2D/3D similarity search and molecular docking with the Vina scoring system.

Peptidomimetics for CVD screened via TRADD-TRAF2 complex interface assessments

The main aim of this study is to screen and develop Peptidomimetics to treat atherosclerosis (AS) which is a Cardio Vascular Disease (CVD). Peptidomimetics were obtained from the protein-protein interaction interface of TRADD (Tumor necrosis factor receptor type 1-associated DEATH domain protein) and TRAF2 (TNF receptor-associated factor 2) complex. TRADD-TRAF2 interaction is critical in AS pathogenesis since it assists a series of signal transducers that activate NF-κB. Conceptually, the triggered NF-κB makes an extensive amount of nitric oxide (NO) synthesized by inducible nitric oxide synthase (iNOS), which boons the progress of AS. The examined TRADD-TRAF2 complex (PDB ID: 1F3V) and its interaction details revealed that the sequence range W11-G165 of TRADD highly interacts with TRAF2. The sequence range W11-G165 was selected for the design and preparation of the inhibitory peptide in silico. The selected sequence was mutated with the alanine scanning method to have a range of inhibitory peptides. With the help of different in silico tools, the top three, namely MIP11-25 L, MIP131-143 h, and MIP149-164 m peptides showed the best interaction with the critical residues of TRAF2. Thus, these three peptides were used for generating peptidomimetics using pepMMsMIMIC, a peptidomimetics virtual screening tool. Around 600 peptidomimetics were identified & and retrieved for further screening by employing molecular docking tools and MD (Molecular Dynamics) simulations. Density Functional Theory (DFT) and ADMET predictions were applied to validate the screened peptidomimetics druggability. In the results, peptidomimic compounds MMs03918858 and MMs03927281 with binding energy values of -9.6 kcal/mol and - 9.1 kcal/mol respectively were screened as the best and are proposed for further pre-clinical studies.

The collision-induced dissociation mechanism of sodiated Hex-HexNAc disaccharides

Determining carbohydrate structures, such as their compositions, linkage positions, and in particular the anomers and stereoisomers, is a great challenge. Isomers of different anomers or stereoisomers have the same sequences of chemical bonds, but have different orientations of some chemical bonds which are difficult to be distinguished by mass spectrometry. Collision-induced dissociation (CID) tandem mass spectroscopy (MS/MS) is a widely used technique for characterizing carbohydrate structures. Understanding the carbohydrate dissociation mechanism is important for obtaining the structural information from MS/MS. In this work, we studied the CID mechanism of galactose-N-acetylgalactosamine (Gal-GalNAc) and glucose-N-acetylglucosamine (Glc-GlcNAc) disaccharides with 1→3 and 1→4 linkages. For Gal-GalNAc disaccharides, the CID mass spectra of sodium ion adducts show significant difference between the α- and β-anomers of GalNAc at the reducing end, while no difference in the CID mass spectra between two anomers of Glc-GlcNAc disaccharides was found. Quantum chemistry calculations show that for Gal-GalNAc disaccharides, the difference of the dissociation barriers between dehydration and glycosidic bond cleavage is significantly small in the β-anomer compared to that in the α-anomer; while these differences are similar between the α- and β-anomers of Glc-GlcNAc disaccharides. These differences can be attributed to the different orientations of hydroxyl and N-acetyl groups located at GalNAc and GlcNAc. The calculation results are consistent with the CID spectra of isotope labelled disaccharides. Our study provides an insight into the CID of 1→3 and 1→4 linked Gal-GalNAc and Glc-GlcNAc disaccharides. This information is useful for determining the anomeric configurations of GalNAc in oligosaccharides.

Identifying and profiling structural similarities between Spike of SARS-CoV-2 and other viral or host proteins with Machaon

Using protein structure to predict function, interactions, and evolutionary history is still an open challenge, with existing approaches relying extensively on protein homology and families. Here, we present Machaon, a data-driven method combining orientation invariant metrics on phi-psi angles, inter-residue contacts and surface complexity. It can be readily applied on whole structures or segments-such as domains and binding sites. Machaon was applied on SARS-CoV-2 Spike monomers of native, Delta and Omicron variants and identified correlations with a wide range of viral proteins from close to distant taxonomy ranks, as well as host proteins, such as ACE2 receptor. Machaon's meta-analysis of the results highlights structural, chemical and transcriptional similarities between the Spike monomer and human proteins, indicating a multi-level viral mimicry. This extended analysis also revealed relationships of the Spike protein with biological processes such as ubiquitination and angiogenesis and highlighted different patterns in virus attachment among the studied variants. Available at: https://machaonweb.com .

Machine Learning Prediction of Hydration Free Energy with Physically Inspired Descriptors

We present machine learning models for predicting experimental hydration free energies of molecules without any atom-, bond-, or geometry-specific input feature. Four types of physically inspired descriptors are adopted for predictions. The first type is composed of the total dipole moment, anisotropic polarizability, and vibrational analysis results of the solute molecule. The second and third types are derived from the electrostatic potential distribution of the solute. The last type includes the solvent accessible surface area and shape similarities. Several machine learning regression models are built on the basis of the FreeSolv database with ∼600 samples, showing a better performance in comparison with that of most traditional approaches and other prediction methods based on molecular fingerprints. In particular, the present descriptors are capable of predicting hydration free energies of new compounds with elements or fragments that are never seen in the training set. The importance of these descriptors, the impact of dissociation energies of specific covalent bonds, and the outliers with relatively large prediction errors are also discussed.

Discovery of two-dimensional binary nanoparticle superlattices using global Monte Carlo optimization

Binary nanoparticle (NP) superlattices exhibit distinct collective plasmonic, magnetic, optical, and electronic properties. Here, we computationally demonstrate how fluid-fluid interfaces could be used to self-assemble binary systems of NPs into 2D superlattices when the NP species exhibit different miscibility with the fluids forming the interface. We develop a basin-hopping Monte Carlo (BHMC) algorithm tailored for interface-trapped structures to rapidly determine the ground-state configuration of NPs, allowing us to explore the repertoire of binary NP architectures formed at the interface. By varying the NP size ratio, interparticle interaction strength, and difference in NP miscibility with the two fluids, we demonstrate the assembly of an array of exquisite 2D periodic architectures, including AB-, AB₂-, and AB₃-type monolayer superlattices as well as AB-, AB₂-, A₃B₅-, and A₄B₆-type bilayer superlattices. Our results suggest that the interfacial assembly approach could be a versatile platform for fabricating 2D colloidal superlattices with tunable structure and properties.

A multi-reference poly-conformational method for in silico design, optimization, and repositioning of pharmaceutical compounds illustrated for selected SARS-CoV-2 ligands

Background

This work presents a novel computational multi-reference poly-conformational algorithm for design, optimization, and repositioning of pharmaceutical compounds.

Methods

The algorithm searches for candidates by comparing similarities between conformers of the same compound and identifies target compounds, whose conformers are collectively close to the conformers of each compound in the reference set. Reference compounds may possess highly variable MoAs, which directly, and simultaneously, shape the properties of target candidate compounds.

Results

The algorithm functionality has been case study validated in silico, by scoring ChEMBL drugs against FDA-approved reference compounds that either have the highest predicted binding affinity to our chosen SARS-CoV-2 targets or are confirmed to be inhibiting such targets in-vivo. All our top scoring ChEMBL compounds also turned out to be either high-affinity ligands to the chosen targets (as confirmed in separate studies) or show significant efficacy, in-vivo, against those selected targets. In addition to method case study validation, in silico search for new compounds within two virtual libraries from the Enamine database is presented. The library's virtual compounds have been compared to the same set of reference drugs that we used for case study validation: Olaparib, Tadalafil, Ergotamine and Remdesivir. The large reference set of four potential SARS-CoV-2 compounds has been selected, since no drug has been identified to be 100% effective against the virus so far, possibly because each candidate drug was targeting only one, particular MoA. The goal here was to introduce a new methodology for identifying potential candidate(s) that cover multiple MoA-s presented within a set of reference compounds.

Fewer Dimensions, More Structures for Improved Discrete Models of Dynamics of Free versus Antigen-Bound Antibody

Over the past decade, Markov State Models (MSM) have emerged as powerful methodologies to build discrete models of dynamics over structures obtained from Molecular Dynamics trajectories. The identification of macrostates for the MSM is a central decision that impacts the quality of the MSM but depends on both the selected representation of a structure and the clustering algorithm utilized over the featurized structures. Motivated by a large molecular system in its free and bound state, this paper investigates two directions of research, further reducing the representation dimensionality in a non-parametric, data-driven manner and including more structures in the computation. Rigorous evaluation of the quality of obtained MSMs via various statistical tests in a comparative setting firmly shows that fewer dimensions and more structures result in a better MSM. Many interesting findings emerge from the best MSM, advancing our understanding of the relationship between antibody dynamics and antibody–antigen recognition.

Maximizing the Performance of Similarity-Based Virtual Screening Methods by Generating Synergy from the Integration of 2D and 3D Approaches

Methods for the pairwise comparison of 2D and 3D molecular structures are established approaches in virtual screening. In this work, we explored three strategies for maximizing the virtual screening performance of these methods: (i) the merging of hit lists obtained from multi-compound screening using a single screening method, (ii) the merging of the hit lists obtained from 2D and 3D screening by parallel selection, and (iii) the combination of both of these strategies in an integrated approach. We found that any of these strategies led to a boost in virtual screening performance, with the clearest advantages observed for the integrated approach. On test sets for virtual screening, covering 50 pharmaceutically relevant proteins, the integrated approach, using sets of five query molecules, yielded, on average, an area under the receiver operating characteristic curve (AUC) of 0.84, an early enrichment among the top 1% of ranked compounds (EF1%) of 53.82 and a scaffold recovery rate among the top 1% of ranked compounds (SRR1%) of 0.50. In comparison, the 2D and 3D methods on their own (when using a single query molecule) yielded AUC values of 0.68 and 0.54, EF1% values of 19.96 and 17.52, and SRR1% values of 0.20 and 0.17, respectively. In conclusion, based on these results, the integration of 2D and 3D methods, via a (balanced) parallel selection strategy, is recommended, and, in particular, when combined with multi-query screening.

Metadynamics molecular dynamics and isothermal Brownian-type molecular dynamics simulations for the chiral cluster Au18

In an effort to gain insight into enantiomeric transitions, their transition mechanism, time span of transitions and distribution of time spans etc, we performed molecular dynamics (MD) simulations on chiral clusters Au₁₀, Au₁₅and Au₁₈, and found that viable reaction coordinates can be deduced from simulation data for enlightening the enantiomeric dynamics for Au₁₀and Au₁₅, but not so for Au₁₈. The failure in translating the Au₁₈-L ⇌ Au₁₈-R transitions by MD simulations has been chalked up to the thermal energyk_BTat 300 K being much lower than energy barriers separating the enantiomers of Au₁₈. Two simulation strategies were taken to resolve this simulation impediment. The first one uses the well-tempered metadynamics MD (MMD) simulation, and the second one adeptly applies first a somewhat crude MMD simulation to locate a highly symmetrical isomer Au_18Sand subsequently employed it as initial configuration in the MD simulation. In both strategies, we work in collective variable space of lower dimensionality. The well-tempered MMD simulation tactic was carried out aiming to offer a direct verification of Au₁₈enantiomers, while the tactic to conduct MMD/MD simulations in two consecutive simulation steps was intended to provide an indirect evidence of the existence of enantiomers of Au₁₈given that energy barriers separating them are much higher than ca.k_BTat 300 K. This second tactic, in addition to confirming indirectly Au₁₈-L and Au₁₈-R starting from the symmetrical cluster Au_18S, the simulation results shed light also on the mechanism akin to associative/nonassociative reaction transitions.

KiSSim: Predicting Off-Targets from Structural Similarities in the Kinome

Protein kinases are among the most important drug targets because their dysregulation can cause cancer, inflammatory and degenerative diseases, and many more. Developing selective inhibitors is challenging due to the highly conserved binding sites across the roughly 500 human kinases. Thus, detecting subtle similarities on a structural level can help explain and predict off-targets among the kinase family. Here, we present the kinase-focused, subpocket-enhanced KiSSim fingerprint (Kinase Structural Similarity). The fingerprint builds on the KLIFS pocket definition, composed of 85 residues aligned across all available protein kinase structures, which enables residue-by-residue comparison without a computationally expensive alignment. The residues' physicochemical and spatial properties are encoded within their structural context including key subpockets at the hinge region, the DFG motif, and the front pocket. Since structure was found to contain information complementary to sequence, we used the fingerprint to calculate all-against-all similarities within the structurally covered kinome. We could identify off-targets that are unexpected if solely considering the sequence-based kinome tree grouping; for example, Erlobinib's known kinase off-targets SLK and LOK show high similarities to the key target EGFR (TK group), although belonging to the STE group. KiSSim reflects profiling data better or at least as well as other approaches such as KLIFS pocket sequence identity, KLIFS interaction fingerprints (IFPs), or SiteAlign. To rationalize observed (dis)similarities, the fingerprint values can be visualized in 3D by coloring structures with residue and feature resolution. We believe that the KiSSim fingerprint is a valuable addition to the kinase research toolbox to guide off-target and polypharmacology prediction. The method is distributed as an open-source Python package on GitHub and as a conda package: https://github.com/volkamerlab/kissim.

Evaluation of Deep Learning Architectures for Aqueous Solubility Prediction

Determining the aqueous solubility of molecules is a vital step in many pharmaceutical, environmental, and energy storage applications. Despite efforts made over decades, there are still challenges associated with developing a solubility prediction model with satisfactory accuracy for many of these applications. The goals of this study are to assess current deep learning methods for solubility prediction, develop a general model capable of predicting the solubility of a broad range of organic molecules, and to understand the impact of data properties, molecular representation, and modeling architecture on predictive performance. Using the largest currently available solubility data set, we implement deep learning-based models to predict solubility from the molecular structure and explore several different molecular representations including molecular descriptors, simplified molecular-input line-entry system strings, molecular graphs, and three-dimensional atomic coordinates using four different neural network architectures-fully connected neural networks, recurrent neural networks, graph neural networks (GNNs), and SchNet. We find that models using molecular descriptors achieve the best performance, with GNN models also achieving good performance. We perform extensive error analysis to understand the molecular properties that influence model performance, perform feature analysis to understand which information about the molecular structure is most valuable for prediction, and perform a transfer learning and data size study to understand the impact of data availability on model performance.

Indexing Ultrafast Shape-Based Descriptors in MongoDB to Identify TLR4 Pathway Agonists

A method is presented for an ultrafast shape-based search workflow for the screening of large compound collections, i.e., those of vendors. The three-dimensional shape of a molecule dictates its biological activity by enabling the molecule to fit into binding pockets of proteins. Quite often, distinctly different chemical compounds that have similar shapes can bind in a similar way. OpenEye pioneered an algorithm for comparing shapes of molecules by overlaying them in a computer and measuring differences between a query molecule and a target molecule. Overlaying shapes is a computationally intensive process and represents a bottleneck in searching for similar molecules. More recent publications describe alternative methods of overlaying molecules, which are accomplished by comparing shape-based descriptors. These methods were implemented in the Open Drug Discovery Toolkit (ODDT) package. We utilized a combination of open-source software packages like ODDT and RDkit to implement a workflow for ultrafast conformer generation and matching that does not require storing precomputed conformers on the file system or in memory. Moreover, the generated descriptors could be optionally stored in MongoDB for performing searches in the future. To speed up the search, we created a set of indexes from the transformed shape-based descriptors. We are in the process of calculating descriptors for multiple vendors, including Enamine's "REAL" collection of 1.2 billion compounds. Currently, the shape similarity search on more than 70 million compounds takes less than 8 s! We exemplified our methodology with the screen of compounds that can act as putative TLR4 agonists. The search was based on a literature-known small-molecule TLR4 agonist series. In due course, we identified compounds with novel structural motifs that were active in mouse and human TLR4 reporter cell lines.

Identifying SARS-CoV-2 main protease inhibitors by applying the computer screening of a large database of molecules

The outbreak of coronavirus disease 2019 (COVID-19) at the end of 2019 affected global health. Its infection agent was called severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Wearing a mask, maintaining social distance, and vaccination are effective ways to prevent infection of SARS-CoV-2, but none of them help infected people. Targeting the enzymes of SARS-CoV-2 is an effective way to stop the replication of the virus in infected people and treat COVID-19 patients. SARS-CoV-2 main protease is a therapeutic target which the inhibition of its enzymatic activity prevents from the replication of SARS-CoV-2. A large database of molecules has been searched to identify new inhibitors for SARS-CoV-2 main protease enzyme. At the first step, ligand screening based on similarity search was used to select similar compounds to known SARS-CoV-2 main protease inhibitors. Then molecules with better predicted pharmacokinetic properties were selected. Structure-based virtual screening based on the application of molecular docking and molecular dynamics simulation methods was used to select more effective inhibitors among selected molecules in previous step. Finally two compounds were considered as SARS-CoV-2 main protease inhibitors.

Amplifying the Sensitivity of Polydiacetylene Sensors: The Dummy Molecule Approach

There is an increasing need for fast and accurate assessment of various health conditions, where polydiacetylenes (PDA), having unique stress-sensitive optical properties, have great potential. When the conjugated backbone of PDA is disturbed by steric repulsion between the receptor-target complexes formed at the PDA surface via specific recognition events, the bandgap of PDA increases and produces color change and fluorescent emission as a dual sensory signal. However, this detection mechanism suggests an intrinsic sensitivity limit of PDA platform because unless adjacent receptors are occupied by target molecules no signal is anticipated. A novel approach to improve the sensitivity and limit of detection of PDA sensors has been developed by preoccupying the surface of PDA liposomes with an optimized amount of artificial target molecules named as dummy molecules. The sensitivity and limit of detection (LOD) showed large improvement by the surface-bound dummy molecules. In addition, the dummy strategy was synergically integrated with another sensitivity enhancing method with a different working mechanism in a PDA sensor for Neomycin detection. When optimized, the LOD of the PDA sensor was improved to 7 nM from 80 nM of the control and the signal intensity increased consistently throughout the entire tested concentration range of the target Neomycin. Finally, the general applicability of the dummy strategy to other target molecules was successfully confirmed by implementing the dummy strategy in a PDA sensor for Surfactin detection.

DeepUMQA: ultrafast shape recognition-based protein model quality assessment using deep learning

MOTIVATION

Protein model quality assessment is a key component of protein structure prediction. In recent research, the voxelization feature was used to characterize the local structural information of residues, but it may be insufficient for describing residue-level topological information. Design features that can further reflect residue-level topology when combined with deep learning methods are therefore crucial to improve the performance of model quality assessment.

RESULTS

We developed a deep-learning method, DeepUMQA, based on Ultrafast Shape Recognition (USR) for the residue-level single-model quality assessment. In the framework of the deep residual neural network, the residue-level USR feature was introduced to describe the topological relationship between the residue and overall structure by calculating the first moment of a set of residue distance sets and then combined with 1D, 2D and voxelization features to assess the quality of the model. Experimental results on the CASP13, CASP14 test datasets and CAMEO blind test show that USR could supplement the voxelization features to comprehensively characterize residue structure information and significantly improve model assessment accuracy. The performance of DeepUMQA ranks among the top during the state-of-the-art single-model quality assessment methods, including ProQ2, ProQ3, ProQ3D, Ornate, VoroMQA, ProteinGCN, ResNetQA, QDeep, GraphQA, ModFOLD6, ModFOLD7, ModFOLD8, QMEAN3, QMEANDisCo3 and DeepAccNet.

AVAILABILITY AND IMPLEMENTATION

The DeepUMQA server is freely available at http://zhanglab-bioinf.com/DeepUMQA/.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Collision-Induced Dissociation of Cellobiose and Maltose

Structure determination is a longstanding bottleneck of carbohydrate research. Tandem mass spectrometry (MS/MS) is one of the most widely used methods for carbohydrate structure determination. However, the effectiveness of MS/MS depends on how the precursor structures are derived from the observed fragments. Understanding the dissociation mechanisms is crucial for MS/MS-based structure determination. Herein, we investigate the collision-induced dissociation mechanism of β-cellobiose and β-maltose sodium adducts using quantum chemical calculations and experimental measurements. Four dissociation channels are studied. Dehydration mainly occurs through the transfer of an H atom to O1 of the sugar at the reducing end, followed by a C1-O1 bond cleavage; cross-ring dissociation starts with a ring-opening reaction, which occurs through the transfer of an H atom from O1 to O5 of the sugar at the reducing end. These two dissociation channels are analogous to that of glucose monosaccharide. The third channel, generation of B₁ and Y₁ ions, occurs through the transfer of an H atom from O3 (cellobiose) or O2 (maltose) to O1 of the sugar at the nonreducing end, followed by a glycosidic bond cleavage. The fourth channel, C₁-Z₁ fragmentation, has two mechanisms: (1) the transfer of an H atom from O3 or O2 to O4 of the sugar at the reducing end to generate C ions in the ring form and (2) the transfer of an H atom from O3 of the sugar at the reducing end to O5 of the sugar at the nonreducing end to produce C ions in the linear form. The results of calculations are supported by experimental collision-induced dissociation spectral measurements.

Using Stochastic Computing for Virtual Screening Acceleration

Stochastic computing is an emerging scientific field pushed by the need for developing high-performance artificial intelligence systems in hardware to quickly solve complex data processing problems. This is the case of virtual screening, a computational task aimed at searching across huge molecular databases for new drug leads. In this work, we show a classification framework in which molecules are described by an energy-based vector. This vector is then processed by an ultra-fast artificial neural network implemented through FPGA by using stochastic computing techniques. Compared to other previously published virtual screening methods, this proposal provides similar or higher accuracy, while it improves processing speed by about two or three orders of magnitude.

Search for Global Minimum Structures of P 2 n + 1 + (n = 1-15) Using xTB-Based Basin-Hopping Algorithm

A new program for searching global minimum structures of atomic clusters using basin-hopping algorithm based on the xTB method was developed here. The program can be performed with a much higher speed than its replacement directly based on DFT methods. Considering the structural varieties and complexities in finding their global minimum structures, phosphorus cluster cations were studied by the program. The global minimum structures of cationic P2n+1+ (n = 1–15) clusters are determined through the unbiased structure searching method. In the last step, further DFT optimization was performed for the selected isomers. For P2n+1+ (n = 1–4, 7), the found global minimum structures are in consistent with the ones previously reported; while for P2n+1+ (n = 5, 6, 8–12), newly found isomers are more energy-favorable than those previously reported. And those for P2n+1+ (n = 13–15) are reported here for the first time. Among them, the most stable isomers of P2n+1+ (n = 4–6, 9) are characterized by their C_3v, C_s, C_2v and C_s symmetry, in turn. But those of P2n+1+ (n = 7, 8, 10–12), no symmetry has been identified. The most stable isomers of P29+ and P31+ are characterized by single P-P bonds bridging units inside the clusters. Further analysis shows that the pnicogen bonds play an important role in the stabilization of these clusters. These results show that the new developed program is effective and robust in searching global minimum structures for atom clusters, and it also provides new insights into the role of pnicogen bonds in phosphorus clusters.

Drugs repurposed for COVID-19 by virtual screening of 6,218 drugs and cell-based assay

Recent spread of SARS-CoV-2 has sparked significant health concerns of emerging infectious viruses. Drug repurposing is a tangible strategy for developing antiviral agents within a short period. In general, drug repurposing starts with virtual screening of approved drugs employing docking simulations. However, the actual hit rate is low, and most of the predicted compounds are false positives. To tackle the challenges, we report advanced virtual screening with pre- and postdocking pharmacophore filtering of 6,218 drugs for COVID-19. Notably, 7 out of 38 compounds showed efficacies in inhibiting SARS-CoV-2 in Vero cells. Three of these were also found to inhibit SARS-CoV-2 in human Calu-3 cells. Furthermore, three drug combinations showed strong synergistic effects in SARS-CoV-2 inhibition at their clinically achievable concentrations.

The COVID-19 pandemic caused by SARS-CoV-2 is an unprecedentedly significant health threat, prompting the need for rapidly developing antiviral drugs for the treatment. Drug repurposing is currently one of the most tangible options for rapidly developing drugs for emerging and reemerging viruses. In general, drug repurposing starts with virtual screening of approved drugs employing various computational methods. However, the actual hit rate of virtual screening is very low, and most of the predicted compounds are false positives. Here, we developed a strategy for virtual screening with much reduced false positives through incorporating predocking filtering based on shape similarity and postdocking filtering based on interaction similarity. We applied this advanced virtual screening approach to repurpose 6,218 approved and clinical trial drugs for COVID-19. All 6,218 compounds were screened against main protease and RNA-dependent RNA polymerase of SARS-CoV-2, resulting in 15 and 23 potential repurposed drugs, respectively. Among them, seven compounds can inhibit SARS-CoV-2 replication in Vero cells. Three of these drugs, emodin, omipalisib, and tipifarnib, show anti-SARS-CoV-2 activities in human lung cells, Calu-3. Notably, the activity of omipalisib is 200-fold higher than that of remdesivir in Calu-3. Furthermore, three drug combinations, omipalisib/remdesivir, tipifarnib/omipalisib, and tipifarnib/remdesivir, show strong synergistic effects in inhibiting SARS-CoV-2. Such drug combination therapy improves antiviral efficacy in SARS-CoV-2 infection and reduces the risk of each drug’s toxicity. The drug repurposing strategy reported here will be useful for rapidly developing drugs for treating COVID-19 and other viruses.

Identification of Anomericity and Linkage of Arabinose and Ribose through Collision-Induced Dissociation

Arabinose and ribose are two common pentoses that exist in both furanose and pyranose forms in plant and bacteria oligosaccharides. In this study, each pentose isomer, namely α-furanose, β-furanose, α-pyranose, and β-pyranose, was first separated through high-performance liquid chromatography followed by an investigation of collision-induced dissociation in an ion trap mass spectrometer. The major dissociation channels, dehydration and cross-ring dissociation, were analyzed by using high-level quantum chemistry calculations and transition state theory. The branching ratio of major dissociation channels was governed by two geometrical features: one being the cis or trans configuration of O1 and O2 atoms determining dehydration preferability and the other being the number of hydroxyl groups on the same side of the ring as the O1 atom determining the preferability of cross-ring dissociation. The relative branching ratios of the major channels were used to identify anomericity and the linkages of arabinose and ribose. Arabinose in the β-configuration and ribose in the α-configuration are predicted to have larger relative dehydration branching ratios than arabinose in the α-configuration and ribose in the β-configuration, respectively. Arabinose and ribose at the reducing end of oligosaccharides with 1 → 2 (pyranose and furanose), 1 → 3 (pyranose and furanose), 1 → 4 (pyranose only), and 1 → 5 (furanose only) linkages are predicted to undergo ^0,2X, ^0,3X, ^0,2A, and ^0,2A/^0,3A cross-ring dissociation, respectively. Application of the dissociation mechanism to the disaccharide linkage determination is demonstrated.

Neural Network Methodology for the Identification and Classification of Lipopeptides Based on SMILES Annotation

Artificial Neural Networks can be applied for the identification and classification of prospective drug candidates such as complex compounds, including lipopeptide, based on their SMILES string representation. The training of neural networks is done with SMILES strings, which are predictive of structural identification; the ANNs are efficient of correctly classifying all compounds, substructures and their analogues distinguishing the drugs based upon atomic organization to obtain lead optimization in drug discovery. The proficiency of the trained ANN models in recognizing and classifying the analogous compounds was tested for analysis of similar compounds, which were not taken previously for training and achieved results with correct classification in the validation set. The best result was achieved with 10 numbers of hidden layers. The R2 value for training is 0.90586; the R2 value for testing is 0.99508; the R2 value after validation is 0.94151; the final value of R2 for total sets is 0.89456. The graphs are plotted between 21 epochs and mean square error (MSE) to report the performance of the model. The value of 798.1735 for the gradient of the curve after 21 iterations and 6 validation checks was obtained. A successful model was developed for the identification and classification of lipopeptides from their SMILES annotation that efficiently classifies similar compounds and supports in decision making for analogue-based drug discovery. This will help in appropriate lead optimization studies for the prediction of potential anticancer and antimicrobial lipopeptide-based therapeutics.

Potential Biological Targets Prediction, ADME Profiling, & Molecular Docking studies of Novel Steroidal Products from Cunninghamella Blakesleana

BACKGROUND

New potential biological targets prediction through inverse molecular docking technique is an another smart strategy to forecast the possibility of compounds being biologically active against various target receptors.

OBJECTIVES

In this case of designed study, we screened our recently obtained novel acetylinic steroidal biotransformed products [(1) 8-β-methyl-14-α-hydroxy∆4tibolone (2) 9-α-Hydroxy∆4 tibolone (3) 8-β-methyl-11-β-hydroxy∆4tibolone (4) 6-β-hydroxy∆4tibolone, (5) 6-β-9-α-dihydroxy∆4tibolone (6) 7-β-hydroxy∆4tibolone) ] from fungi Cunninghemella Blakesleana to predict their possible biological targets and profiling of ADME properties.

METHOD

The prediction of pharmacokinetics properties membrane permeability as well as bioavailability radar properties were carried out by using Swiss target prediction, and Swiss ADME tools, respectively these metabolites were also subjected to predict the possible mechanism of action along with associated biological network pathways by using Reactome data-base.

RESULTS

All the six screened compounds possess excellent drug ability criteria, and exhibited exceptionally excellent non inhibitory potential against all five isozymes of CYP450 enzyme complex, including (CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4) respectively. All the screened compounds are lying within the acceptable pink zone of bioavailability radar and showing excellent descriptive properties. Compounds [1-4 & 6] are showing high BBB (Blood Brain Barrier) permeation, while compound 5 is exhibiting high HIA (Human Intestinal Absorption) property of (Egan Egg).

CONCLUSION

In conclusion, the results of this study smartly reveals that in-silico based studies are considered to provide robustness towards a rational drug designing and development approach, therefore in this way it helps to avoid the possibility of failure of drug candidates in the later experimental stages of drug development phases.

MolAICal: a soft tool for 3D drug design of protein targets by artificial intelligence and classical algorithm

Deep learning is an important branch of artificial intelligence that has been successfully applied into medicine and two-dimensional ligand design. The three-dimensional (3D) ligand generation in the 3D pocket of protein target is an interesting and challenging issue for drug design by deep learning. Here, the MolAICal software is introduced to supply a way for generating 3D drugs in the 3D pocket of protein targets by combining with merits of deep learning model and classical algorithm. The MolAICal software mainly contains two modules for 3D drug design. In the first module of MolAICal, it employs the genetic algorithm, deep learning model trained by FDA-approved drug fragments and Vinardo score fitting on the basis of PDBbind database for drug design. In the second module, it uses deep learning generative model trained by drug-like molecules of ZINC database and molecular docking invoked by Autodock Vina automatically. Besides, the Lipinski's rule of five, Pan-assay interference compounds (PAINS), synthetic accessibility (SA) and other user-defined rules are introduced for filtering out unwanted ligands in MolAICal. To show the drug design modules of MolAICal, the membrane protein glucagon receptor and non-membrane protein SARS-CoV-2 main protease are chosen as the investigative drug targets. The results show MolAICal can generate the various and novel ligands with good binding scores and appropriate XLOGP values. We believe that MolAICal can use the advantages of deep learning model and classical programming for designing 3D drugs in protein pocket. MolAICal is freely for any nonprofit purpose and accessible at https://molaical.github.io.

Drug Repurposing: An Emerging Tool for Drug Reuse, Recycling and Discovery

Drug repositioning or repurposing is a revolutionary breakthrough in drug development that focuses on rediscovering new uses for old therapeutic agents. Drug repositioning can be defined more precisely as the process of exploring new indications for an already approved drug while drug repurposing includes overall re-development approaches grounded in the identical chemical structure of the active drug moiety as in the original product. The repositioning approach accelerates the drug development process, curtails the cost and risk inherent to drug development. The strategy focuses on the polypharmacology of drugs to unlocks novel opportunities for logically designing more efficient therapeutic agents for unmet medical disorders. Drug repositioning also expresses certain regulatory challenges that hamper its further utilization. The review outlines the eminent role of drug repositioning in new drug discovery, methods to predict the molecular targets of a drug molecule, advantages that the strategy offers to the pharmaceutical industries, explaining how the industrial collaborations with academics can assist in the discovering more repositioning opportunities. The focus of the review is to highlight the latest applications of drug repositioning in various disorders. The review also includes a comparison of old and new therapeutic uses of repurposed drugs, assessing their novel mechanisms of action and pharmacological effects in the management of various disorders. Various restrictions and challenges that repurposed drugs come across during their development and regulatory phases are also highlighted.

Collision-induced dissociation of xylose and its applications in linkage and anomericity identification

Different dehydration barrier heights result in different branching ratio, a simple and fast anomeric configuration identification for xylose.

A pair potential modeling study of F3− in neon matrices

First-principles investigations of the trifluoride anion in a neon environment reveal a small blue-shift of the fundamental vibrational excitations.

Nanoparticles Incorporating a Fluorescence Turn-on Reporter for Real-Time Drug Release Monitoring, a Chemoenhancer and a Stealth Agent: Poseidon's Trident against Cancer?

The rate and extent of drug release under physiological conditions is a key factor influencing the therapeutic activity of a formulation. Real-time detection of drug release by conventional pharmacokinetics approaches is confounded by low sensitivity, particularly in the case of tissue-targeted novel drug delivery systems, where low concentrations of the drug reach systemic circulation. We present a novel fluorescence turn-on platform for real-time monitoring of drug release from nanoparticles based on reversible fluorescence quenching in fluorescein esters. Fluorescein-conjugated carbon nanotubes (CNTs) were esterified with methotrexate in solution and solid phase, followed by supramolecular functionalization with a chemoenhancer (suramin) or/and a stealth agent (dextran sulfate). Suramin was found to increase the cytotoxicity of methotrexate in A549 cells. On the other hand, dextran sulfate exhibited no effect on cytotoxicity or cellular uptake of CNTs by A549 cells, while a decrease in cellular uptake of CNTs and cytotoxicity of methotrexate was observed in macrophages (RAW 264.7 cells). Similar results were also obtained when CNTs were replaced with graphene. Docking studies revealed that the conjugates are not internalized by folate receptors/transporters. Further, docking and molecular dynamics studies revealed the conjugates do not exhibit affinity toward the methotrexate target, dihydrofolate reductase. Molecular dynamics studies also revealed that distinct features of dextran-CNT and suramin-CNT interactions, characterized by π-π interactions between CNTs and dextran/suramin. Our study provides a simple, cost-effective, and scalable method for the synthesis of nanoparticles conferred with the ability to monitor drug release in real-time. This method could also be extended to other drugs and other types of nanoparticles.

Enantiomeric Transitions in the Chiral Cluster Au15 Studied by a Reaction Coordinate Deduced from Molecular Dynamics Simulations

A recently developed modified basin hopping (MBH) optimization algorithm, combined with an energy function calculated by the semiempirical density functional tight-binding (DFTB) theory, was applied to determine the lowest-energy structures of Au_n clusters with size n = 3-20. It was predicted from the DFTB/MBH optimization algorithm calculations that clusters Au₁₀, Au₁₅, and Au₁₈ exhibit chiral properties; i.e., each of these three clusters possesses the same energy value and associated with it are two nonsuperposable mirror-image clusters. In the potential energy landscape, there thus exist multidimensional barriers separating the two enantiomers, and this lowest-energy double-well morphology is surrounded by potential-energy minima of higher energies. In this paper, we have chosen to study the chiral cluster Au₁₅ by employing an isothermal Brownian-type molecular dynamics simulation to discern in greater detail its conformational transition from one enantiomer, say left, to its right counterpart. To facilitate our analysis of the simulation data, we transpose the multidimensional configurational space description to a lower dimensional collective variable (CV) space spanned by two geometry-relevant CVs. The thermally driven progression and mechanism of enantiomeric transitions between the left and right enantiomers will be our main focus, and the strategy is to dissect the time development of the CVs collected from different sets of independent simulation runs. From simulation data, we found that an understanding of the dynamics of enantiomeric transitions needs first to seek out the left and right enantiomers through a molecular modeling and visualizing program, then to ferret out and identify between the left and right enantiomers a symmetrical structure, and finally to define from the latter a reaction coordinate. We showed in this work that this single reaction coordinate is predictive in unraveling the left ⇌ right enantiomeric transition events, providing a specific inkling of the transition time span and its associated distribution which can be checked further for its reasonableness by the autocorrelation function and a vibrational analysis, all of which shed light on the mechanisms of transition.

A method for predicting basins in the global optimization of nanoclusters with applications to AlxCuy alloys

The problem of obtaining the geometrical configuration of a molecule that minimizes its potential energy is a very complicated one for a series of applications, ranging from determining the structure of biological macromolecules to nanoclusters of atoms. Global optimization tools are available for this task, and many of them are based in performing successive local optimizations, where the starting geometries for these steps are determined by an intelligent algorithm. Here we develop a method to save computing time in the optimization of nanoclusters by predicting if a given minimum has been previously visited during local optimization steps. Our application to Cu-Al nanoalloys indicates that it is possible to save a substantial amount of computational cost. The application also reveals new promising Al_xCu_y clusters and explain their stabilities in terms of the jellium model.

sensaas: Shape-based Alignment by Registration of Colored Point-based Surfaces

sensaas is a tool developed for aligning and comparing molecular shapes and sub-shapes. Alignment is obtained by registration of 3D point-based representations of the van der Waals surface. The method uses local properties of the shape to identify the correspondence relationships between two point clouds containing up to several thousand colored (labeled) points. Our rigid-body superimposition method follows a two-stage approach. An initial alignment is obtained by matching pose-invariant local 3D descriptors, called FPFH, of the input point clouds. This stage provides a global superimposition of the molecular surfaces, without any knowledge of their initial pose in 3D space. This alignment is then refined by optimizing the matching of colored points. In our study, each point is colored according to its closest atom, which itself belongs to a user defined physico-chemical class. Finally, sensaas provides an alignment and evaluates the molecular similarity by using Tversky coefficients. To assess the efficiency of this approach, we tested its ability to reproduce the superimposition of X-ray structures of the benchmarking AstraZeneca (AZ) data set and, compared its results with those generated by the two shape-alignment approaches shaep and shafts. We also illustrated submatching properties of our method with respect to few substructures and bioisosteric fragments. The code is available upon request from the authors (demo version at https://chemoinfo.ipmc.cnrs.fr/SENSAAS).

Real time structural search of the Protein Data Bank

Detection of protein structure similarity is a central challenge in structural bioinformatics. Comparisons are usually performed at the polypeptide chain level, however the functional form of a protein within the cell is often an oligomer. This fact, together with recent growth of oligomeric structures in the Protein Data Bank (PDB), demands more efficient approaches to oligomeric assembly alignment/retrieval. Traditional methods use atom level information, which can be complicated by the presence of topological permutations within a polypeptide chain and/or subunit rearrangements. These challenges can be overcome by comparing electron density volumes directly. But, brute force alignment of 3D data is a compute intensive search problem. We developed a 3D Zernike moment normalization procedure to orient electron density volumes and assess similarity with unprecedented speed. Similarity searching with this approach enables real-time retrieval of proteins/protein assemblies resembling a target, from PDB or user input, together with resulting alignments (http://shape.rcsb.org).

TF3P: Three-Dimensional Force Fields Fingerprint Learned by Deep Capsular Network

Molecular fingerprints are the workhorse in ligand-based drug discovery. In recent years, an increasing number of research papers reported fascinating results on using deep neural networks to learn 2D molecular representations as fingerprints. It is anticipated that the integration of deep learning would also contribute to the prosperity of 3D fingerprints. Here, we unprecedentedly introduce deep learning into 3D small molecule fingerprints, presenting a new one we termed as the three-dimensional force fields fingerprint (TF3P). TF3P is learned by a deep capsular network whose training is in no need of labeled data sets for specific predictive tasks. TF3P can encode the 3D force fields information of molecules and demonstrates the stronger ability to capture 3D structural changes, to recognize molecules alike in 3D but not in 2D, and to identify similar targets inaccessible by other 2D or 3D fingerprints based on only ligands similarity. Furthermore, TF3P is compatible with both statistical models (e.g., similarity ensemble approach) and machine learning models. Altogether, we report TF3P as a new 3D small molecule fingerprint with a promising future in ligand-based drug discovery. All codes are written in Python and available at https://github.com/canisw/tf3p.

Reducing Ensembles of Protein Tertiary Structures Generated De Novo via Clustering

Controlling the quality of tertiary structures computed for a protein molecule remains a central challenge in de-novo protein structure prediction. The rule of thumb is to generate as many structures as can be afforded, effectively acknowledging that having more structures increases the likelihood that some will reside near the sought biologically-active structure. A major drawback with this approach is that computing a large number of structures imposes time and space costs. In this paper, we propose a novel clustering-based approach which we demonstrate to significantly reduce an ensemble of generated structures without sacrificing quality. Evaluations are related on both benchmark and CASP target proteins. Structure ensembles subjected to the proposed approach and the source code of the proposed approach are publicly-available at the links provided in Section 1.

Structure- and Ligand-Based Virtual Screening on DUD-E+: Performance Dependence on Approximations to the Binding Pocket

Using the DUD-E⁺ benchmark, we explore the impact of using a single protein pocket or ligand for virtual screening compared with using ensembles of alternative pockets, ligands, and sets thereof. For both structure-based and ligand-based approaches, the precise characterization of the binding site in question had a significant impact on screening performance. Using the single original DUD-E protein, Surflex-Dock yielded mean ROC area of 0.81 ± 0.11. Using the cognate ligand instead, with the eSim method for screening, yielded 0.77 ± 0.14. Moving to ensembles of five protein pocket variants increased docking performance to 0.84 ± 0.09. Results for the analogous ligand-based approach (using the five crystallographically aligned cognate ligands) was 0.83 ± 0.11. Using the same ligands, but making use of an automatically generated mutual alignment, yielded mean AUC nearly as good as from single-structure docking: 0.80 ± 0.12. Detailed results and statistical analyses show that structure- and ligand-based methods are complementary and can be fruitfully combined to enhance screening efficiency. A hybrid approach combining ensemble docking with eSim-based screening produced the best and most consistent performance (mean ROC area of 0.89 ± 0.08 and 1% early enrichment of 46-fold). Based on results from both the docking and ligand-similarity approaches, it is clearly unwise to make use of a single arbitrarily chosen protein structure for docking or single ligand query for similarity-based screening.

Comprehensive investigation of selectivity landscape of glycogen synthase kinase-3 inhibitors

Interaction signatures of drug candidates are characteristic to off-target (neutral) and antitarget (negative) effects, inferring reduced efficiency, side-effects and high attrition rate. Today's retroactive scaled-down virtual screening (VS) experiments relying on benchmarking datasets are extensively involved to assess ligand enrichment in the real-world problem. In recent years, unbiased benchmarking sets turned into a tremendous need to assist virtual screening methodologies for emerging drug targets. To date, the benchmarking datasets are quite limited, whereas glycogen synthase kinase-3 (GSK-3) is not included into directories of benchmarking datasets such as DUD-e, MUV, etc. Herein we introduced our in-house algorithm to build an unbiased benchmarking dataset, including highly selective, moderately selective and nonselective inhibitors for a significant therapeutic target - GSK-3, suitable for both ligand-based and structure-based VS approaches. These datasets are unbiased in terms of physico-chemical properties and topological descriptors, as resulted from mean(ROC-AUC) leave-one-out cross-validation (LOO CV). and additional 2 D similarity search. Moreover, we investigated the gradual selectivity dataset by application of multiple 2 D similarity coefficients and distances, 3 D similarity and docking. Besides the resulted links between the enrichment of selective GSK-3 inhibitors and their chemical structures, a database of compounds and their 3 D similarity signatures including cut-off thresholds for enhanced selectivity was generated. 2 D similarity space analysis revealed that selectivity problem cannot be evaluated appropriately with 2 D similarity searching alone. The current analysis provided useful, comprehensive insights, which may facilitate the knowledge-based identification of novel selective GSK-3 inhibitors.Communicated by Ramaswamy H. Sarma.

Shape Similarity by Fractal Dimensionality: An Application in the de novo Design of (-)-Englerin A Mimetics

Molecular shape and pharmacological function are interconnected. To capture shape, the fractal dimensionality concept was employed, providing a natural similarity measure for the virtual screening of de novo generated small molecules mimicking the structurally complex natural product (-)-englerin A. Two of the top-ranking designs were synthesized and tested for their ability to modulate transient receptor potential (TRP) cation channels which are cellular targets of (-)-englerin A. Intracellular calcium assays and electrophysiological whole-cell measurements of TRPC4 and TRPM8 channels revealed potent inhibitory effects of one of the computer-generated compounds. Four derivatives of this identified hit compound had comparable effects on TRPC4 and TRPM8. The results of this study corroborate the use of fractal dimensionality as an innovative shape-based molecular representation for molecular scaffold-hopping.

Applying Machine Learning to Ultrafast Shape Recognition in Ligand-Based Virtual Screening

Ultrafast Shape Recognition (USR), along with its derivatives, are Ligand-Based Virtual Screening (LBVS) methods that condense 3-dimensional information about molecular shape, as well as other properties, into a small set of numeric descriptors. These can be used to efficiently compute a measure of similarity between pairs of molecules using a simple inverse Manhattan Distance metric. In this study we explore the use of suitable Machine Learning techniques that can be trained using USR descriptors, so as to improve the similarity detection of potential new leads. We use molecules from the Directory for Useful Decoys-Enhanced to construct machine learning models based on three different algorithms: Gaussian Mixture Models (GMMs), Isolation Forests and Artificial Neural Networks (ANNs). We train models based on full molecule conformer models, as well as the Lowest Energy Conformations (LECs) only. We also investigate the performance of our models when trained on smaller datasets so as to model virtual screening scenarios when only a small number of actives are known a priori. Our results indicate significant performance gains over a state of the art USR-derived method, ElectroShape 5D, with GMMs obtaining a mean performance up to 430% better than that of ElectroShape 5D in terms of Enrichment Factor with a maximum improvement of up to 940%. Additionally, we demonstrate that our models are capable of maintaining their performance, in terms of enrichment factor, within 10% of the mean as the size of the training dataset is successively reduced. Furthermore, we also demonstrate that running times for retrospective screening using the machine learning models we selected are faster than standard USR, on average by a factor of 10, including the time required for training. Our results show that machine learning techniques can significantly improve the virtual screening performance and efficiency of the USR family of methods.

Building maps of protein structure spaces in template-free protein structure prediction

An important goal in template-free protein structure prediction is how to control the quality of computed tertiary structures of a target amino-acid sequence. Despite great advances in algorithmic research, given the size, dimensionality, and inherent characteristics of the protein structure space, this task remains exceptionally challenging. It is current practice to aim to generate as many structures as can be afforded so as to increase the likelihood that some of them will reside near the sought but unknown biologically-active/native structure. When operating within a given computational budget, this is impractical and uninformed by any metrics of interest. In this paper, we propose instead to equip algorithms that generate tertiary structures, also known as decoy generation algorithms, with memory of the protein structure space that they explore. Specifically, we propose an evolving, granularity-controllable map of the protein structure space that makes use of low-dimensional representations of protein structures. Evaluations on diverse target sequences that include recent hard CASP targets show that drastic reductions in storage can be made without sacrificing decoy quality. The presented results make the case that integrating a map of the protein structure space is a promising mechanism to enhance decoy generation algorithms in template-free protein structure prediction.