Application scenario-oriented molecule generation platform developed for drug discovery.

Design of molecules for candidate compound selection is one of the central challenges in drug discovery due to the complexity of chemical space and requirement of multi-parameter optimization. Here we present an application scenario-oriented platform (ID4Idea) for molecule generation in different scenarios of drug discovery. This platform utilizes both library or rule based and generative based algorithms (VAE, RNN, GAN, etc.), in combination with various AI learning types (pre-training, transfer learning, reinforcement learning, active learning, etc.) and input representations (1D SMILES, 2D graph, 3D shape, binding site, pharmacophore, etc.), to enable customized solutions for a given molecular design scenario. Besides the usual generation followed screening protocol, goal-directed molecule generation can also be conducted towards predefined goals, enhancing the efficiency of hit identification, lead finding, and lead optimization. We demonstrate the effectiveness of ID4Idea platform through case studies, showcasing customized solutions for different design tasks using various input information, such as binding pockets, pharmacophores, and compound representations. In addition, remaining challenges are discussed to unlock the full potential of AI models in drug discovery and pave the way for the development of novel therapeutics.

A Combination of Machine Learning and PBPK Modeling Approach for Pharmacokinetics Prediction of Small Molecules in Humans.

PURPOSE: Recently, there has been rapid development in model-informed drug development, which has the potential to reduce animal experiments and accelerate drug discovery. Physiologically based pharmacokinetic (PBPK) and machine learning (ML) models are commonly used in early drug discovery to predict drug properties. However, basic PBPK models require a large number of molecule-specific inputs from in vitro experiments, which hinders the efficiency and accuracy of these models. To address this issue, this paper introduces a new computational platform that combines ML and PBPK models. The platform predicts molecule PK profiles with high accuracy and without the need for experimental data. METHODS: This study developed a whole-body PBPK model and ML models of plasma protein fraction unbound ( f up ), Caco-2 cell permeability, and total plasma clearance to predict the PK of small molecules after intravenous administration. Pharmacokinetic profiles were simulated using a "bottom-up" PBPK modeling approach with ML inputs. Additionally, 40 compounds were used to evaluate the platform's accuracy. RESULTS: Results showed that the ML-PBPK model predicted the area under the concentration-time curve (AUC) with 65.0 % accuracy within a 2-fold range, which was higher than using in vitro inputs with 47.5 % accuracy. CONCLUSION: The ML-PBPK model platform provides high accuracy in prediction and reduces the number of experiments and time required compared to traditional PBPK approaches. The platform successfully predicts human PK parameters without in vitro and in vivo experiments and can potentially guide early drug discovery and development.

Systematic Studies on the Protocol and Criteria for Selecting a Covalent Docking Tool.

With the resurgence of drugs with covalent binding mechanisms, much attention has been paid to docking methods for the discovery of targeted covalent inhibitors. The existence of many available covalent docking tools has inspired development of a systematic and objective procedure and criteria with which to evaluate these programs. In order to find a tool appropriate to studies of a covalently binding system, protocols and criteria are proposed for protein-ligand covalent docking studies. This paper consists of three sections: (1) curating a standard data set to evaluate covalent docking tools objectively; (2) establishing criteria to measure the performance of a tool applied for docking ligands into a complex system; and (3) creating a protocol to evaluate and select covalent binding tools. The protocols were applied to evaluate four covalent docking tools (MOE, GOLD, CovDock, and ICM-Pro) and parameters affecting covalent docking performance were investigated.

cBinderDB: a covalent binding agent database.

Motivation: Small molecule drug candidates with attractive toxicity profiles that modulate target proteins through non-covalent interactions are usually favored by scientists and pharmaceutical industry. In the past decades, many non-covalent binding agents have been developed for different diseases. However, an increasing attention has been paid to covalent binding agents in pharmaceutical fields during recent years. Many covalent binding agents entered clinical trials and exerted significant advantages for diseases such as infection, cancers, gastrointestinal disorders, central nervous system or cardiovascular diseases. It has been recognized that covalent binding ligands can be attractive drug candidates. But, there is lack of resource to support covalent ligand discovery. Results: Hence, we initiated a covalent binder database (cBinderDB). To our best knowledge, it is the first online database that provides information on covalent binding compound structures, chemotypes, targets, covalent binding types and other biological properties. The covalent binding targets are annotated with biological functions, protein family and domains, gene information, modulators and receptor-ligand complex structure. The data in the database were collected from scientific publications by combining a text mining method and manual inspection processes. cBinderDB covers covalent binder's data up to September 2016. Availability and Implementation: cBinderDB is freely available at www.rcdd.org.cn/cbinderdb/. Contact: guqiong@mail.sysu.edu.cn or junxu@biochemomes.com . Supplementary information: Supplementary data are available at Bioinformatics online.

TCMAnalyzer: A Chemo- and Bioinformatics Web Service for Analyzing Traditional Chinese Medicine.

Traditional Chinese medicine (TCM) has been widely used and proven effective in long term clinical practice. However, the molecular mechanism of action for many TCMs remains unclear due to the complexity of many ingredients and their interactions with biological receptors. This is one of the major roadblocks in TCM modernization. In order to solve this problem, we have developed TCMAnalyzer, which is a free web-based toolkit allowing a user to (1) identify the potential compounds that are responsible for the bioactivities for a TCM herb through scaffold-activity relation searches using structural search techniques, (2) investigate the molecular mechanism of action for a TCM herb at the systemic level, and (3) explore the potentially targeted bioactive herbs. The toolkit can result in TCM networks that demonstrate the relations among natural product molecules (small molecular ligands), putative protein targets, pathways, and diseases. These networks are graphically depicted to reveal the mechanism of actions for a TCM herb or to identify new molecular scaffolds for new chemotherapies. TCMAnalyzer is freely available at http://www.rcdd.org.cn/tcmanalyzer .

Predicting DPP-IV inhibitors with machine learning approaches.

Dipeptidyl peptidase IV (DPP-IV) is a promising Type 2 diabetes mellitus (T2DM) drug target. DPP-IV inhibitors prolong the action of glucagon-like peptide-1 (GLP-1) and gastric inhibitory peptide (GIP), improve glucose homeostasis without weight gain, edema, and hypoglycemia. However, the marketed DPP-IV inhibitors have adverse effects such as nasopharyngitis, headache, nausea, hypersensitivity, skin reactions and pancreatitis. Therefore, it is still expected for novel DPP-IV inhibitors with minimal adverse effects. The scaffolds of existing DPP-IV inhibitors are structurally diversified. This makes it difficult to build virtual screening models based upon the known DPP-IV inhibitor libraries using conventional QSAR approaches. In this paper, we report a new strategy to predict DPP-IV inhibitors with machine learning approaches involving naïve Bayesian (NB) and recursive partitioning (RP) methods. We built 247 machine learning models based on 1307 known DPP-IV inhibitors with optimized molecular properties and topological fingerprints as descriptors. The overall predictive accuracies of the optimized models were greater than 80%. An external test set, composed of 65 recently reported compounds, was employed to validate the optimized models. The results demonstrated that both NB and RP models have a good predictive ability based on different combinations of descriptors. Twenty "good" and twenty "bad" structural fragments for DPP-IV inhibitors can also be derived from these models for inspiring the new DPP-IV inhibitor scaffold design.

Integrating FRAM and BN for enhanced resilience evaluation in construction emergency response: A scaffold collapse case study.

The construction system's complexity can generate substantial uncertainties during emergencies. Resilience, as a new perspective on emergency response, can significantly mitigate these challenges. This paper introduces an innovative model to assess the resilience of construction emergency response processes utilizing a scaffold collapse scenario as a demonstrative case study. Grounded in resilience engineering, our model integrates the merits of the Functional Resonance Analysis Method (FRAM) with the probabilistic strengths of Bayesian Networks (BNs). The process commences with FRAM, mapping out the emergency response in qualitative terms by identifying functions, variabilities, and couplings. This culminates in a topological network which serves as a foundational structure for the directed Complex Network (CN) and the BN model. Thereafter, the Delphi method and the modified K-shell (MKS) decomposition algorithm guide the computation of prior probabilities for root nodes and the conditional probability table within the BN model. Subsequently, the BN model is subjected to a simulation using the AgenaRisk software, executing both forward and backward propagation as well as sensitivity analyses. Our findings pinpoint "Intersectoral Coordination and Linkage" as the most crucial function, with rapidity being the most sensitive aspect influencing resilience during a scaffold collapse emergency response process.

DenovoProfiling: A webserver for de novo generated molecule library profiling.

Various deep learning-based architectures for molecular generation have been proposed for de novo drug design. The flourish of the de novo molecular generation methods and applications has created a great demand for the visualization and functional profiling for the de novo generated molecules. An increasing number of publicly available chemogenomic databases sets good foundations and creates good opportunities for comprehensive profiling of the de novo library. In this paper, we present DenovoProfiling, a webserver dedicated to de novo library visualization and functional profiling. Currently, DenovoProfiling contains six modules: (1) identification & visualization module for chemical structure visualization and identify the reported structures, (2) chemical space module for chemical space exploration using similarity maps, principal components analysis (PCA), drug-like properties distribution, and scaffold-based clustering, (3) ADMET prediction module for predicting the ADMET properties of the de novo molecules, (4) molecular alignment module for three dimensional molecular shape analysis, (5) drugs mapping module for identifying structural similar drugs, and (6) target & pathway module for identifying the reported targets and corresponding functional pathways. DenovoProfiling could provide structural identification, chemical space exploration, drug mapping, and target & pathway information. The comprehensive annotated information could give users a clear picture of their de novo library and could guide the further selection of candidates for chemical synthesis and biological confirmation. DenovoProfiling is freely available at http://denovoprofiling.xielab.net.

TCMIO: A Comprehensive Database of Traditional Chinese Medicine on Immuno-Oncology.

Advances in immuno-oncology (IO) are making immunotherapy a powerful tool for cancer treatment. With the discovery of an increasing number of IO targets, many herbs or ingredients from traditional Chinese medicine (TCM) have shown immunomodulatory function and antitumor effects via targeting the immune system. However, knowledge of underlying mechanisms is limited due to the complexity of TCM, which has multiple ingredients acting on multiple targets. To address this issue, we present TCMIO, a comprehensive database of Traditional Chinese Medicine on Immuno-Oncology, which can be used to explore the molecular mechanisms of TCM in modulating the cancer immune microenvironment. Over 120,000 small molecules against 400 IO targets were extracted from public databases and the literature. These ligands were further mapped to the chemical ingredients of TCM to identify herbs that interact with the IO targets. Furthermore, we applied a network inference-based approach to identify the potential IO targets of natural products in TCM. All of these data, along with cheminformatics and bioinformatics tools, were integrated into the publicly accessible database. Chemical structure mining tools are provided to explore the chemical ingredients and ligands against IO targets. Herb-ingredient-target networks can be generated online, and pathway enrichment analysis for TCM or prescription is available. This database is functional for chemical ingredient structure mining and network analysis for TCM. We believe that this database provides a comprehensive resource for further research on the exploration of the mechanisms of TCM in cancer immunity and TCM-inspired identification of novel drug leads for cancer immunotherapy. TCMIO can be publicly accessed at http://tcmio.xielab.net.

DeepScreening: a deep learning-based screening web server for accelerating drug discovery.

Deep learning contributes significantly to researches in biological sciences and drug discovery. Previous studies suggested that deep learning techniques have shown superior performance to other machine learning algorithms in virtual screening, which is a critical step to accelerate the drug discovery. However, the application of deep learning techniques in drug discovery and chemical biology are hindered due to the data availability, data further processing and lacking of the user-friendly deep learning tools and interface. Therefore, we developed a user-friendly web server with integration of the state of art deep learning algorithm, which utilizes either the public or user-provided dataset to help biologists or chemists perform virtual screening either the chemical probes or drugs for a specific target of interest. With DeepScreening, user could conveniently construct a deep learning model and generate the target-focused de novo libraries. The constructed classification and regression models could be subsequently used for virtual screening against the generated de novo libraries, or diverse chemical libraries in stock. From deep models training to virtual screening, and target focused de novo library generation, all those tasks could be finished with DeepScreening. We believe this deep learning-based web server will benefit to both biologists and chemists for probes or drugs discovery.

PTS: a pharmaceutical target seeker.

Identifying protein targets for a bioactive compound is critical in drug discovery. Molecular similarity is a main approach to fish drug targets, and is based upon an axiom that similar compounds may have the same targets. The molecular structural similarity of a compound and the ligand of a known target can be gauged in topological (2D), steric (3D) or static (pharmacophoric) metric. The topologic metric is fast, but unable to represent steric and static profile of a bioactive compound. Steric and static metrics reflect the shape properties of a compound if its structure were experimentally obtained, and could be unreliable if they were based upon the putative conformation data. In this paper, we report a pharmaceutical target seeker (PTS), which searches protein targets for a bioactive compound based upon the static and steric shape comparison by comparing a compound structure against the experimental ligand structure. Especially, the crystal structures of active compounds were taken into similarity calculation and the predicted targets can be filtered according to multi activity thresholds. PTS has a pharmaceutical target database that contains approximately 250 000 ligands annotated with about 2300 protein targets. A visualization tool is provided for a user to examine the result. Database URL: http://www.rcdd.org.cn/PTS.