Enhancing property and activity prediction and interpretation using multiple molecular graph representations with MMGX.

Graph Neural Networks (GNNs) excel in compound property and activity prediction, but the choice of molecular graph representations significantly influences model learning and interpretation. While atom-level molecular graphs resemble natural topology, they overlook key substructures or functional groups and their interpretation partially aligns with chemical intuition. Recent research suggests alternative representations using reduced molecular graphs to integrate higher-level chemical information and leverages both representations for model. However, there is a lack of studies about applicability and impact of different molecular graphs on model learning and interpretation. Here, we introduce MMGX (Multiple Molecular Graph eXplainable discovery), investigating the effects of multiple molecular graphs, including Atom, Pharmacophore, JunctionTree, and FunctionalGroup, on model learning and interpretation with various perspectives. Our findings indicate that multiple graphs relatively improve model performance, but in varying degrees depending on datasets. Interpretation from multiple graphs in different views provides more comprehensive features and potential substructures consistent with background knowledge. These results help to understand model decisions and offer valuable insights for subsequent tasks. The concept of multiple molecular graph representations and diverse interpretation perspectives has broad applicability across tasks, architectures, and explanation techniques, enhancing model learning and interpretation for relevant applications in drug discovery.

SpatialPPI: Three-dimensional space protein-protein interaction prediction with AlphaFold Multimer.

Rapid advancements in protein sequencing technology have resulted in gaps between proteins with identified sequences and those with mapped structures. Although sequence-based predictions offer insights, they can be incomplete due to the absence of structural details. Conversely, structure-based methods face challenges with respect to newly sequenced proteins. The AlphaFold Multimer has remarkable accuracy in predicting the structure of protein complexes. However, it cannot distinguish whether the input protein sequences can interact. Nonetheless, by analyzing the information in the models predicted by the AlphaFold Multimer, we propose a highly accurate method for predicting protein interactions. This study focuses on the use of deep neural networks, specifically to analyze protein complex structures predicted by the AlphaFold Multimer. By transforming atomic coordinates and utilizing sophisticated image-processing techniques, vital 3D structural details were extracted from protein complexes. Recognizing the significance of evaluating residue distances in protein interactions, this study leveraged image recognition approaches by integrating Densely Connected Convolutional Networks (DenseNet) and Deep Residual Network (ResNet) within 3D convolutional networks for protein 3D structure analysis. When benchmarked against leading protein-protein interaction prediction methods, such as SpeedPPI, D-script, DeepTrio, and PEPPI, our proposed method, named SpatialPPI, exhibited notable efficacy, emphasizing the promising role of 3D spatial processing in advancing the realm of structural biology. The SpatialPPI code is available at: https://github.com/ohuelab/SpatialPPI.

Variational autoencoder-based chemical latent space for large molecular structures with 3D complexity.

The structural diversity of chemical libraries, which are systematic collections of compounds that have potential to bind to biomolecules, can be represented by chemical latent space. A chemical latent space is a projection of a compound structure into a mathematical space based on several molecular features, and it can express structural diversity within a compound library in order to explore a broader chemical space and generate novel compound structures for drug candidates. In this study, we developed a deep-learning method, called NP-VAE (Natural Product-oriented Variational Autoencoder), based on variational autoencoder for managing hard-to-analyze datasets from DrugBank and large molecular structures such as natural compounds with chirality, an essential factor in the 3D complexity of compounds. NP-VAE was successful in constructing the chemical latent space from large-sized compounds that were unable to be handled in existing methods, achieving higher reconstruction accuracy, and demonstrating stable performance as a generative model across various indices. Furthermore, by exploring the acquired latent space, we succeeded in comprehensively analyzing a compound library containing natural compounds and generating novel compound structures with optimized functions.

Design of Cyclic Peptides Targeting Protein-Protein Interactions Using AlphaFold.

More than 930,000 protein-protein interactions (PPIs) have been identified in recent years, but their physicochemical properties differ from conventional drug targets, complicating the use of conventional small molecules as modalities. Cyclic peptides are a promising modality for targeting PPIs, but it is difficult to predict the structure of a target protein-cyclic peptide complex or to design a cyclic peptide sequence that binds to the target protein using computational methods. Recently, AlphaFold with a cyclic offset has enabled predicting the structure of cyclic peptides, thereby enabling de novo cyclic peptide designs. We developed a cyclic peptide complex offset to enable the structural prediction of target proteins and cyclic peptide complexes and found AlphaFold2 with a cyclic peptide complex offset can predict structures with high accuracy. We also applied the cyclic peptide complex offset to the binder hallucination protocol of AfDesign, a de novo protein design method using AlphaFold, and we could design a high predicted local-distance difference test and lower separated binding energy per unit interface area than the native MDM2/p53 structure. Furthermore, the method was applied to 12 other protein-peptide complexes and one protein-protein complex. Our approach shows that it is possible to design putative cyclic peptide sequences targeting PPI.

MEGADOCK-on-Colab: an easy-to-use protein-protein docking tool on Google Colaboratory.

MOTIVATION: Since the advent of ColabFold, numerous software packages have been provided with Google Colaboratory-compatible ipynb files, allowing users to effortlessly test and reproduce results without the need for local installation or configuration. MEGADOCK, a protein-protein docking tool, is particularly well-suited for Google Colaboratory due to its lightweight computations and GPU acceleration capabilities. To increase accessibility and promote widespread use, it is crucial to provide a computing environment compatible with Google Colaboratory. RESULTS: In this study, we report the development of a Google Colaboratory environment for running our protein-protein docking software, MEGADOCK. We provide a comprehensive ipynb file, including the compilation of MEGADOCK with the FFTW library installation on Colaboratory, the introduction of related tools using PyPI/apt, and the execution and visualization of docking structures. This streamlined environment enables users to visualize docking structures with just one click. The code is available under a CC-BY NC 4.0 license from https://github.com/ohuelab/MEGADOCK-on-Colab .

Generating Potential Protein-Protein Interaction Inhibitor Molecules Based on Physicochemical Properties.

Protein-protein interactions (PPIs) are associated with various diseases; hence, they are important targets in drug discovery. However, the physicochemical empirical properties of PPI-targeted drugs are distinct from those of conventional small molecule oral pharmaceuticals, which adhere to the "rule of five (RO5)". Therefore, developing PPI-targeted drugs using conventional methods, such as molecular generation models, is challenging. In this study, we propose a molecular generation model based on deep reinforcement learning that is specialized for the production of PPI inhibitors. By introducing a scoring function that can represent the properties of PPI inhibitors, we successfully generated potential PPI inhibitor compounds. These newly constructed virtual compounds possess the desired properties for PPI inhibitors, and they show similarity to commercially available PPI libraries. The virtual compounds are freely available as a virtual library.

CycPeptMPDB: A Comprehensive Database of Membrane Permeability of Cyclic Peptides.

Recently, cyclic peptides have been considered breakthrough drugs because they can interact with "undruggable" targets such as intracellular protein-protein interactions. Membrane permeability is an essential indicator of oral bioavailability and intracellular targeting, and the development of membrane-permeable peptides is a bottleneck in cyclic peptide drug discovery. Although many experimental data on membrane permeability of cyclic peptides have been reported, a comprehensive database is not yet available. A comprehensive membrane permeability database is essential for developing computational methods for cyclic peptide drug design. In this study, we constructed CycPeptMPDB, the first web-accessible database of cyclic peptide membrane permeability. We collected information on a total of 7334 cyclic peptides, including the structure and experimentally measured membrane permeability, from 45 published papers and 2 patents from pharmaceutical companies. To unambiguously represent cyclic peptides larger than small molecules, we used the hierarchical editing language for macromolecules notation to generate a uniform sequence representation of peptides. In addition to data storage, CycPeptMPDB provides several supporting functions such as online data visualization, data analysis, and downloading. CycPeptMPDB is expected to be a valuable platform to support membrane permeability research on cyclic peptides. CycPeptMPDB can be freely accessed at http://cycpeptmpdb.com.

Effective Protein-Ligand Docking Strategy via Fragment Reuse and a Proof-of-Concept Implementation.

Virtual screening is a commonly used process to search for feasible drug candidates from a huge number of compounds during the early stages of drug design. As the compound database continues to expand to billions of entries or more, there remains an urgent need to accelerate the process of docking calculations. Reuse of calculation results is a possible way to accelerate the process. In this study, we first propose yet another virtual screening-oriented docking strategy by combining three factors, namely, compound decomposition, simplified fragment grid storing k-best scores, and flexibility consideration with pregenerated conformers. Candidate compounds contain many common fragments (chemical substructures). Thus, the calculation results of these common fragments can be reused among them. As a proof-of-concept of the aforementioned strategies, we also conducted the development of REstretto, a tool that implements the three factors to enable the reuse of calculation results. We demonstrated that the speed and accuracy of REstretto were comparable to those of AutoDock Vina, a well-known free docking tool. The implementation of REstretto has much room for further performance improvement, and therefore, the results show the feasibility of the strategy. The code is available under an MIT license at https://github.com/akiyamalab/restretto.

Lipid Composition Is Critical for Accurate Membrane Permeability Prediction of Cyclic Peptides by Molecular Dynamics Simulations.

Cyclic peptides have attracted attention as a promising pharmaceutical modality due to their potential to selectively inhibit previously undruggable targets, such as intracellular protein-protein interactions. Poor membrane permeability is the biggest bottleneck hindering successful drug discovery based on cyclic peptides. Therefore, the development of computational methods that can predict membrane permeability and support elucidation of the membrane permeation mechanism of drug candidate peptides is much sought after. In this study, we developed a protocol to simulate the behavior in membrane permeation steps and estimate the membrane permeability of large cyclic peptides with more than or equal to 10 residues. This protocol requires the use of a more realistic membrane model than a single-lipid phospholipid bilayer. To select a membrane model, we first analyzed the effect of cholesterol concentration in the model membrane on the potential of mean force and hydrogen bonding networks along the direction perpendicular to the membrane surface as predicted by molecular dynamics simulations using cyclosporine A. These results suggest that a membrane model with 40 or 50 mol % cholesterol was suitable for predicting the permeation process. Subsequently, two types of membrane models containing 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine and 40 and 50 mol % cholesterol were used. To validate the efficiency of our protocol, the membrane permeability of 18 ten-residue peptides was predicted. Correlation coefficients of R > 0.8 between the experimental and calculated permeability values were obtained with both model membranes. The results of this study demonstrate that the lipid membrane is not just a medium but also among the main factors determining the membrane permeability of molecules. The computational protocol proposed in this study and the findings obtained on the effect of membrane model composition will contribute to building a schematic view of the membrane permeation process. Furthermore, the results of this study will eventually aid the elucidation of design rules for peptide drugs with high membrane permeability.

Solubility-Aware Protein Binding Peptide Design Using AlphaFold.

New protein-protein interactions (PPIs) are identified, but PPIs have different physicochemical properties compared with conventional targets, making it difficult to use small molecules. Peptides offer a new modality to target PPIs, but designing appropriate peptide sequences by computation is challenging. Recently, AlphaFold and RoseTTAFold have made it possible to predict protein structures from amino acid sequences with ultra-high accuracy, enabling de novo protein design. We designed peptides likely to have PPI as the target protein using the "binder hallucination" protocol of AfDesign, a de novo protein design method using AlphaFold. However, the solubility of the peptides tended to be low. Therefore, we designed a solubility loss function using solubility indices for amino acids and developed a solubility-aware AfDesign binder hallucination protocol. The peptide solubility in sequences designed using the new protocol increased with the weight of the solubility loss function; moreover, they captured the characteristics of the solubility indices. Moreover, the new protocol sequences tended to have higher affinity than random or single residue substitution sequences when evaluated by docking binding affinity. Our approach shows that it is possible to design peptide sequences that can bind to the interface of PPI while controlling solubility.

Plasma protein binding prediction focusing on residue-level features and circularity of cyclic peptides by deep learning.

MOTIVATION: In recent years, cyclic peptide drugs have been receiving increasing attention because they can target proteins that are difficult to be tackled by conventional small-molecule drugs or antibody drugs. Plasma protein binding rate (%PPB) is a significant pharmacokinetic property of a compound in drug discovery and design. However, due to structural differences, previous computational prediction methods developed for small-molecule compounds cannot be successfully applied to cyclic peptides, and methods for predicting the PPB rate of cyclic peptides with high accuracy are not yet available. RESULTS: Cyclic peptides are larger than small molecules, and their local structures have a considerable impact on PPB; thus, molecular descriptors expressing residue-level local features of cyclic peptides, instead of those expressing the entire molecule, as well as the circularity of the cyclic peptides should be considered. Therefore, we developed a prediction method named CycPeptPPB using deep learning that considers both factors. First, the macrocycle ring of cyclic peptides was decomposed residue by residue. The residue-based descriptors were arranged according to the sequence information of the cyclic peptide. Furthermore, the circular data augmentation method was used, and the circular convolution method CyclicConv was devised to express the cyclic structure. CycPeptPPB exhibited excellent performance, with mean absolute error (MAE) of 4.79% and correlation coefficient (R) of 0.92 for the public drug dataset, compared to the prediction performance of the existing PPB rate prediction software (MAE=15.08%, R=0.63). AVAILABILITY AND IMPLEMENTATION: The data underlying this article are available in the online supplementary material. The source code of CycPeptPPB is available at https://github.com/akiyamalab/cycpeptppb. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Quantitative Estimate Index for Early-Stage Screening of Compounds Targeting Protein-Protein Interactions.

Drug-likeness quantification is useful for screening drug candidates. Quantitative estimates of drug-likeness (QED) are commonly used to assess quantitative drug efficacy but are not suitable for screening compounds targeting protein-protein interactions (PPIs), which have recently gained attention. Therefore, we developed a quantitative estimate index for compounds targeting PPIs (QEPPI), specifically for early-stage screening of PPI-targeting compounds. QEPPI is an extension of the QED method for PPI-targeting drugs that models physicochemical properties based on the information available for drugs/compounds, specifically those reported to act on PPIs. FDA-approved drugs and compounds in iPPI-DB, which comprise PPI inhibitors and stabilizers, were evaluated using QEPPI. The results showed that QEPPI is more suitable than QED for early screening of PPI-targeting compounds. QEPPI was also considered an extended concept of the "Rule-of-Four" (RO4), a PPI inhibitor index. We evaluated the discriminatory performance of QEPPI and RO4 for datasets of PPI-target compounds and FDA-approved drugs using F-score and other indices. The F-scores of RO4 and QEPPI were 0.451 and 0.501, respectively. QEPPI showed better performance and enabled quantification of drug-likeness for early-stage PPI drug discovery. Hence, it can be used as an initial filter to efficiently screen PPI-targeting compounds.

Improved Large-Scale Homology Search by Two-Step Seed Search Using Multiple Reduced Amino Acid Alphabets.

Metagenomic analysis, a technique used to comprehensively analyze microorganisms present in the environment, requires performing high-precision homology searches on large amounts of sequencing data, the size of which has increased dramatically with the development of next-generation sequencing. NCBI BLAST is the most widely used software for performing homology searches, but its speed is insufficient for the throughput of current DNA sequencers. In this paper, we propose a new, high-performance homology search algorithm that employs a two-step seed search strategy using multiple reduced amino acid alphabets to identify highly similar subsequences. Additionally, we evaluated the validity of the proposed method against several existing tools. Our method was faster than any other existing program for ≤120,000 queries, while DIAMOND, an existing tool, was the fastest method for >120,000 queries.

Large-Scale Membrane Permeability Prediction of Cyclic Peptides Crossing a Lipid Bilayer Based on Enhanced Sampling Molecular Dynamics Simulations.

Membrane permeability is a significant obstacle facing the development of cyclic peptide drugs. However, membrane permeation mechanisms are poorly understood. To investigate common features of permeable (and nonpermeable) designs, it is necessary to reproduce the membrane permeation process of cyclic peptides through the lipid bilayer. We simulated the membrane permeation process of 100 six-residue cyclic peptides across the lipid bilayer based on steered molecular dynamics (MD) and replica-exchange umbrella sampling simulations and predicted membrane permeability using the inhomogeneous solubility-diffusion model and a modified version of it. Furthermore, we confirmed the effectiveness of this protocol by predicting the membrane permeability of 56 eight-residue cyclic peptides with diverse chemical structures, including some confidential designs from a pharmaceutical company. As a result, a reasonable correlation between experimentally assessed and calculated membrane permeability of cyclic peptides was observed for the peptide libraries, except for strongly hydrophobic peptides. Our analysis of the MD trajectory demonstrated that most peptides were stabilized in the boundary region between bulk water and membrane and that for most peptides, the process of crossing the center of the membrane is the main obstacle to membrane permeation. The height of this barrier is well correlated with the electrostatic interaction between the peptide and the surrounding media. The structural and energetic features of the representative peptide at each vertical position within the membrane were also analyzed, revealing that peptides permeate the membrane by changing their orientation and conformation according to the surrounding environment.

Taxonomic and Gene Category Analyses of Subgingival Plaques from a Group of Japanese Individuals with and without Periodontitis.

Periodontitis is an inflammation of tooth-supporting tissues, which is caused by bacteria in the subgingival plaque (biofilm) and the host immune response. Traditionally, subgingival pathogens have been investigated using methods such as culturing, DNA probes, or PCR. The development of next-generation sequencing made it possible to investigate the whole microbiome in the subgingival plaque. Previous studies have implicated dysbiosis of the subgingival microbiome in the etiology of periodontitis. However, details are still lacking. In this study, we conducted a metagenomic analysis of subgingival plaque samples from a group of Japanese individuals with and without periodontitis. In the taxonomic composition analysis, genus Bacteroides and Mycobacterium demonstrated significantly different compositions between healthy sites and sites with periodontal pockets. The results from the relative abundance of functional gene categories, carbohydrate metabolism, glycan biosynthesis and metabolism, amino acid metabolism, replication and repair showed significant differences between healthy sites and sites with periodontal pockets. These results provide important insights into the shift in the taxonomic and functional gene category abundance caused by dysbiosis, which occurs during the progression of periodontal disease.

Evaluation of CONSRANK-Like Scoring Functions for Rescoring Ensembles of Protein-Protein Docking Poses.

Scoring is a challenging step in protein-protein docking, where typically thousands of solutions are generated. In this study, we ought to investigate the contribution of consensus-rescoring, as introduced by Oliva et al. (2013) with the CONSRANK method, where the set of solutions is used to build statistics in order to identify recurrent solutions. We explore several ways to perform consensus-based rescoring on the ZDOCK decoy set for Benchmark 4. We show that the information of the interface size is critical for successful rescoring in this context, but that consensus rescoring in itself performs less well than traditional physics-based evaluation. The results of physics-based and consensus-based rescoring are partially overlapping, supporting the use of a combination of these approaches.

Structural Basis for the Binding Mechanism of Human Serum Albumin Complexed with Cyclic Peptide Dalbavancin.

Cyclic peptides, with unique structural features, have emerged as new candidates for drug discovery; their association with human serum albumin (HSA; long blood half-life) is crucial to improve drug delivery and avoid renal clearance. Here, we present the crystal structure of HSA complexed with dalbavancin, a clinically used cyclic peptide. Small-angle X-ray scattering and isothermal titration calorimetry experiments showed that the HSA-dalbavancin complex exists in a monomeric state; dalbavancin is only bound to the subdomain IA of HSA in solution. Structural analysis and MD simulation revealed that the swing of Phe70 and movement of the helix near dalbavancin were necessary for binding. The flip of Leu251 promoted the formation of the binding pocket with an induced-fit mechanism; moreover, the movement of the loop region including Glu60 increased the number of noncovalent interactions with HSA. These findings may support the development of new cyclic peptides for clinical use, particularly the elucidation of their binding mechanism to HSA.

Multidomain protein structure prediction using information about residues interacting on multimeric protein interfaces.

Protein functions can be predicted based on their three-dimensional structures. However, many multidomain proteins have unstable structures, making it difficult to determine the whole structure in biological experiments. Additionally, multidomain proteins are often decomposed and identified based on their domains, with the structure of each domain often found in public databases. Recent studies have advanced structure prediction methods of multidomain proteins through computational analysis. In existing methods, proteins that serve as templates are used for three-dimensional structure prediction. However, when no protein template is available, the accuracy of the prediction is decreased. This study was conducted to predict the structures of multidomain proteins without the need for whole structure templates. We improved structure prediction methods by performing rigid-body docking from the structure of each domain and reranking a structure closer to the correct structure to have a higher value. In the proposed method, the score for the domain-domain interaction obtained without a structural template of the multidomain protein and score for the three-dimensional structure obtained during docking calculation were newly incorporated into the score function. We successfully predicted the structures of 50 of 55 multidomain proteins examined in the test dataset. Interaction residue pair information of the protein-protein complex interface contributes to domain reorganizations even when a structural template for a multidomain protein cannot be obtained. This approach may be useful for predicting the structures of multidomain proteins with important biochemical functions.