A human neutralizing antibody targets the receptor-binding site of SARS-CoV-2.

An outbreak of coronavirus disease 2019 (COVID-19)1-3, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)4, has spread globally. Countermeasures are needed to treat and prevent further dissemination of the virus. Here we report the isolation of two specific human monoclonal antibodies (termed CA1 and CB6) from a patient convalescing from COVID-19. CA1 and CB6 demonstrated potent SARS-CoV-2-specific neutralization activity in vitro. In addition, CB6 inhibited infection with SARS-CoV-2 in rhesus monkeys in both prophylactic and treatment settings. We also performed structural studies, which revealed that CB6 recognizes an epitope that overlaps with angiotensin-converting enzyme 2 (ACE2)-binding sites in the SARS-CoV-2 receptor-binding domain, and thereby interferes with virus-receptor interactions by both steric hindrance and direct competition for interface residues. Our results suggest that CB6 deserves further study as a candidate for translation to the clinic.

Identifying potential small molecule-miRNA associations via Robust PCA based on γ-norm regularization.

Dysregulation of microRNAs (miRNAs) is closely associated with refractory human diseases, and the identification of potential associations between small molecule (SM) drugs and miRNAs can provide valuable insights for clinical treatment. Existing computational techniques for inferring potential associations suffer from limitations in terms of accuracy and efficiency. To address these challenges, we devise a novel predictive model called RPCA$\Gamma $NR, in which we propose a new Robust principal component analysis (PCA) framework based on $\gamma $-norm and $l_{2,1}$-norm regularization and design an Augmented Lagrange Multiplier method to optimize it, thereby deriving the association scores. The Gaussian Interaction Profile Kernel Similarity is calculated to capture the similarity information of SMs and miRNAs in known associations. Through extensive evaluation, including Cross Validation Experiments, Independent Validation Experiment, Efficiency Analysis, Ablation Experiment, Matrix Sparsity Analysis, and Case Studies, RPCA$\Gamma $NR outperforms state-of-the-art models concerning accuracy, efficiency and robustness. In conclusion, RPCA$\Gamma $NR can significantly streamline the process of determining SM-miRNA associations, thus contributing to advancements in drug development and disease treatment.

MARPPI: boosting prediction of protein-protein interactions with multi-scale architecture residual network.

Protein-protein interactions (PPIs) are a major component of the cellular biochemical reaction network. Rich sequence information and machine learning techniques reduce the dependence of exploring PPIs on wet experiments, which are costly and time-consuming. This paper proposes a PPI prediction model, multi-scale architecture residual network for PPIs (MARPPI), based on dual-channel and multi-feature. Multi-feature leverages Res2vec to obtain the association information between residues, and utilizes pseudo amino acid composition, autocorrelation descriptors and multivariate mutual information to achieve the amino acid composition and order information, physicochemical properties and information entropy, respectively. Dual channel utilizes multi-scale architecture improved ResNet network which extracts protein sequence features to reduce protein feature loss. Compared with other advanced methods, MARPPI achieves 96.03%, 99.01% and 91.80% accuracy in the intraspecific datasets of Saccharomyces cerevisiae, Human and Helicobacter pylori, respectively. The accuracy on the two interspecific datasets of Human-Bacillus anthracis and Human-Yersinia pestis is 97.29%, and 95.30%, respectively. In addition, results on specific datasets of disease (neurodegenerative and metabolic disorders) demonstrate the ability to detect hidden interactions. To better illustrate the performance of MARPPI, evaluations on independent datasets and PPIs network suggest that MARPPI can be used to predict cross-species interactions. The above shows that MARPPI can be regarded as a concise, efficient and accurate tool for PPI datasets.

MEG-PPIS: a fast protein-protein interaction site prediction method based on multi-scale graph information and equivariant graph neural network.

MOTIVATION: Protein-protein interaction sites (PPIS) are crucial for deciphering protein action mechanisms and related medical research, which is the key issue in protein action research. Recent studies have shown that graph neural networks have achieved outstanding performance in predicting PPIS. However, these studies often neglect the modeling of information at different scales in the graph and the symmetry of protein molecules within three-dimensional space. RESULTS: In response to this gap, this article proposes the MEG-PPIS approach, a PPIS prediction method based on multi-scale graph information and E(n) equivariant graph neural network (EGNN). There are two channels in MEG-PPIS: the original graph and the subgraph obtained by graph pooling. The model can iteratively update the features of the original graph and subgraph through the weight-sharing EGNN. Subsequently, the max-pooling operation aggregates the updated features of the original graph and subgraph. Ultimately, the model feeds node features into the prediction layer to obtain prediction results. Comparative assessments against other methods on benchmark datasets reveal that MEG-PPIS achieves optimal performance across all evaluation metrics and gets the fastest runtime. Furthermore, specific case studies demonstrate that our method can predict more true positive and true negative sites than the current best method, proving that our model achieves better performance in the PPIS prediction task. AVAILABILITY AND IMPLEMENTATION: The data and code are available at https://github.com/dhz234/MEG-PPIS.git.

A genome-wide association study identifies a transporter for zinc uploading to maize kernels.

The Zn content in cereal seeds is an important trait for crop production as well as for human health. However, little is known about how Zn is loaded to plant seeds. Here, through a genome-wide association study (GWAS), we identify the Zn-NA (nicotianamine) transporter gene ZmYSL2 that is responsible for loading Zn to maize kernels. High promoter sequence variation in ZmYSL2 most likely drives the natural variation in Zn concentrations in maize kernels. ZmYSL2 is specifically localized on the plasma membrane facing the maternal tissue of the basal endosperm transfer cell layer (BETL) and functions in loading Zn-NA into the BETL. Overexpression of ZmYSL2 increases the Zn concentration in the kernels by 31.6%, which achieves the goal of Zn biofortification of maize. These findings resolve the mystery underlying the loading of Zn into plant seeds, providing an efficient strategy for breeding or engineering maize varieties with enriched Zn nutrition.

LncRNA LUESCC promotes esophageal squamous cell carcinoma by targeting the miR-6785-5p/NRSN2 axis.

Esophageal squamous cell carcinoma (ESCC) is one of the most prevalent gastrointestinal malignancies with high mortality worldwide. Emerging evidence indicates that long noncoding RNAs (lncRNAs) are involved in human cancers, including ESCC. However, the detailed mechanisms of lncRNAs in the regulation of ESCC progression remain incompletely understood. LUESCC was upregulated in ESCC tissues compared with adjacent normal tissues, which was associated with gender, deep invasion, lymph node metastasis, and poor prognosis of ESCC patients. LUESCC was mainly localized in the cytoplasm of ESCC cells. Knockdown of LUESCC inhibited cell proliferation, colony formation, migration, and invasion in vitro and suppressed tumor growth in vivo. Mechanistic investigation indicated that LUESCC functions as a ceRNA by sponging miR-6785-5p to enhance NRSN2 expression, which is critical for the malignant behaviors of ESCC. Furthermore, ASO targeting LUESCC substantially suppressed ESCC both in vitro and in vivo. Collectively, these data demonstrate that LUESCC may exerts its oncogenic role by sponging miR-6785-5p to promote NRSN2 expression in ESCC, providing a potential diagnostic marker and therapeutic target for ESCC patients.

SPOCK2 modulates neuropathic pain by interacting with MT1-MMP to regulate astrocytic MMP-2 activation in rats with chronic constriction injury.

BACKGROUND: Neuropathic pain (NP) is a kind of intractable pain. The pathogenesis of NP remains a complicated issue for pain management practitioners. SPARC/osteonectin, CWCV, and Kazal-like domains proteoglycan 2 (SPOCK2) are members of the SPOCK family that play a significant role in the development of the central nervous system. In this study, we investigated the role of SPOCK2 in the development of NP in a rat model of chronic constriction injury (CCI). METHODS: Sprague-Dawley rats were randomly grouped to establish CCI models. We examined the effects of SPOCK2 on pain hpersensitivity and spinal astrocyte activation after CCI-induced NP. Paw withdrawal threshold (PWT) and paw withdrawal latency (PWL) were used to reflects the pain behavioral degree. Molecular mechanisms involved in SPOCK2-mediated NP in vivo were examined by western blot analysis, immunofluorescence, immunohistochemistry, and co-immunoprecipitation. In addition, we examined the SPOCK2-mediated potential protein-protein interaction (PPI) in vitro coimmunoprecipitation (Co-IP) experiments. RESULTS: We founded the expression level of SPOCK2 in rat spinal cord was markedly increased after CCI-induced NP, while SPOCK2 downregulation could partially relieve pain caused by CCI. Our research showed that SPOCK2 expressed significantly increase in spinal astrocytes when CCI-induced NP. In addition, SPOCK2 could act as an upstream signaling molecule to regulate the activation of matrix metalloproteinase-2 (MMP-2), thus affecting astrocytic ERK1/2 activation and interleukin (IL)-1ß production in the development of NP. Moreover, in vitro coimmunoprecipitation (Co-IP) experiments showed that SPOCK2 could interact with membrane-type 1 matrix metalloproteinase (MT1-MMP/MMP14) to regulate MMP-2 activation by the SPARC extracellular (SPARC_EC) domain. CONCLUSIONS: Research shows that SPOCK2 can interact with MT1-MMP to regulate MMP-2 activation, thus affecting astrocytic ERK1/2 activation and IL-1ß production to achieve positive promotion of NP.

Regulating Cu Oxidation State for Electrocatalytic CO2 Conversion into CO with Near-Unity Selectivity via Oxygen Spillover.

Cu-based catalysts are the most intensively studied in the field of electrocatalytic CO2 reduction reaction (CO2RR), demonstrating the capacity to yield diverse C1 and C2+ products albeit with unsatisfactory selectivity. Manipulation of the oxidation state of Cu sites during CO2RR process proves advantageous in modulating the selectivity of productions, but poses a formidable challenge. Here, an oxygen spillover strategy is proposed to enhance the oxidation state of Cu during CO2RR by incorporating the oxygen donor Sb2O4. The Cu-Sb bimetallic oxide catalyst attains a remarkable CO2-to-CO selectivity approaching unity, in stark contrast to the diverse product distribution observed with bare CuO. The exceptional Faradaic efficiency of CO can be maintained across a wide range of potential windows of ≈700 mV in 1 m KOH, and remains independent of the Cu/Sb ratio (ranging from 0.1:1 to 10:1). Correlative calculations and experimental results reveal that oxygen spillover from Sb2O4 to Cu sites maintains the relatively high valence state of Cu during CO2RR, which diminishes the binding strength of *CO, thereby achieving heightened selectivity in CO production. These findings propose the role of oxygen spillover in CO2RR over Cu-based catalysts, and shed light on the rational design of highly selective CO2 reduction catalysts.

AMDE: a novel attention-mechanism-based multidimensional feature encoder for drug-drug interaction prediction.

The properties of the drug may be altered by the combination, which may cause unexpected drug-drug interactions (DDIs). Prediction of DDIs provides combination strategies of drugs for systematic and effective treatment. In most of deep learning-based methods for predicting DDI, encoded information about the drugs is insufficient in some extent, which limits the performances of DDIs prediction. In this work, we propose a novel attention-mechanism-based multidimensional feature encoder for DDIs prediction, namely attention-based multidimensional feature encoder (AMDE). Specifically, in AMDE, we encode drug features from multiple dimensions, including information from both Simplified Molecular-Input Line-Entry System sequence and atomic graph of the drug. Data experiments are conducted on DDI data set selected from Drugbank, involving a total of 34 282 DDI relationships with 17 141 positive DDI samples and 17 141 negative samples. Experimental results show that our AMDE performs better than some state-of-the-art baseline methods, including Random Forest, One-Dimension Convolutional Neural Networks, DeepDrug, Long Short-Term Memory, Seq2seq, Deepconv, DeepDDI, Graph Attention Networks and Knowledge Graph Neural Networks. In practice, we select a set of 150 drugs with 3723 DDIs, which are never appeared in training, validation and test sets. AMDE performs well in DDIs prediction task, with AUROC and AUPRC 0.981 and 0.975. As well, we use Torasemide (DB00214) as an example and predict the most likely drug to interact with it. The top 15 scores all have been reported with clear interactions in literatures.

Molecular substructure tree generative model for de novo drug design.

Deep learning shortens the cycle of the drug discovery for its success in extracting features of molecules and proteins. Generating new molecules with deep learning methods could enlarge the molecule space and obtain molecules with specific properties. However, it is also a challenging task considering that the connections between atoms are constrained by chemical rules. Aiming at generating and optimizing new valid molecules, this article proposed Molecular Substructure Tree Generative Model, in which the molecule is generated by adding substructure gradually. The proposed model is based on the Variational Auto-Encoder architecture, which uses the encoder to map molecules to the latent vector space, and then builds an autoregressive generative model as a decoder to generate new molecules from Gaussian distribution. At the same time, for the molecular optimization task, a molecular optimization model based on CycleGAN was constructed. Experiments showed that the model could generate valid and novel molecules, and the optimized model effectively improves the molecular properties.

De novo molecular design with deep molecular generative models for PPI inhibitors.

We construct a protein-protein interaction (PPI) targeted drug-likeness dataset and propose a deep molecular generative framework to generate novel drug-likeness molecules from the features of the seed compounds. This framework gains inspiration from published molecular generative models, uses the key features associated with PPI inhibitors as input and develops deep molecular generative models for de novo molecular design of PPI inhibitors. For the first time, quantitative estimation index for compounds targeting PPI was applied to the evaluation of the molecular generation model for de novo design of PPI-targeted compounds. Our results estimated that the generated molecules had better PPI-targeted drug-likeness and drug-likeness. Additionally, our model also exhibits comparable performance to other several state-of-the-art molecule generation models. The generated molecules share chemical space with iPPI-DB inhibitors as demonstrated by chemical space analysis. The peptide characterization-oriented design of PPI inhibitors and the ligand-based design of PPI inhibitors are explored. Finally, we recommend that this framework will be an important step forward for the de novo design of PPI-targeted therapeutics.

Suppression of DNMT2/3 by proinflammatory cytokines inhibits CtBP1/2-dependent genes to promote the occurrence of atrophic nonunion.

Failure of bone healing after fracture often results in nonunion, but the underlying mechanism of nonunion pathogenesis is poorly understood. Herein, we provide evidence to clarify that the inflammatory microenvironment of atrophic nonunion (AN) mice suppresses the expression levels of DNA methyltransferases 2 (DNMT2) and 3A (DNMT3a), preventing the methylation of CpG islands on the promoters of C-terminal binding protein 1/2 (CtBP1/2) and resulting in their overexpression. Increased CtBP1/2 acts as transcriptional corepressors that, along with histone acetyltransferase p300 and Runt-related transcription factor 2 (Runx2), suppress the expression levels of six genes involved in bone healing: BGLAP (bone gamma-carboxyglutamate protein), ALPL (alkaline phosphatase), SPP1 (secreted phosphoprotein 1), COL1A1 (collagen 1a1), IBSP (integrin binding sialoprotein), and MMP13 (matrix metallopeptidase 13). We also observe a similar phenomenon in osteoblast cells treated with proinflammatory cytokines or treated with a DNMT inhibitor (5-azacytidine). Forced expression of DNMT2/3a or blockage of CtBP1/2 with their inhibitors can reverse the expression levels of BGLAP/ALPL/SPP1/COL1A1/IBSP/MMP13 in the presence of proinflammatory cytokines. Administration of CtBP1/2 inhibitors in fractured mice can prevent the incidence of AN. Thus, we demonstrate that the downregulation of bone healing genes dependent on proinflammatory cytokines/DNMT2/3a/CtBP1/2-p300-Runx2 axis signaling plays a critical role in the pathogenesis of AN. Disruption of this signaling may represent a new therapeutic strategy to prevent AN incidence after bone fracture.

Micropeptides: potential treatment strategies for cancer.

Some noncoding RNAs (ncRNAs) carry open reading frames (ORFs) that can be translated into micropeptides, although noncoding RNAs (ncRNAs) have been previously assumed to constitute a class of RNA transcripts without coding capacity. Furthermore, recent studies have revealed that ncRNA-derived micropeptides exhibit regulatory functions in the development of many tumours. Although some of these micropeptides inhibit tumour growth, others promote it. Understanding the role of ncRNA-encoded micropeptides in cancer poses new challenges for cancer research, but also offers promising prospects for cancer therapy. In this review, we summarize the types of ncRNAs that can encode micropeptides, highlighting recent technical developments that have made it easier to research micropeptides, such as ribosome analysis, mass spectrometry, bioinformatics methods, and CRISPR/Cas9. Furthermore, based on the distribution of micropeptides in different subcellular locations, we explain the biological functions of micropeptides in different human cancers and discuss their underestimated potential as diagnostic biomarkers and anticancer therapeutic targets in clinical applications, information that may contribute to the discovery and development of new micropeptide-based tools for early diagnosis and anticancer drug development.

The Warburg effect drives cachectic states in patients with pancreatobiliary adenocarcinoma.

We have studied whether the Warburg effect (uncontrolled glycolysis) in pancreatobiliary adenocarcinoma triggers cachexia in the patient. After 74 pancreatobiliary adenocarcinomas were removed by surgery, their glucose transporter-1 and four glycolytic enzymes were quantified using Western blotting. Based on the resulting data, the adenocarcinomas were equally divided into a group of low glycolysis (LG) and a group of high glycolysis (HG). Energy homeostasis was assessed in these cancer patients and in 74 non-cancer controls, using serum albumin and C-reactive protein and morphometrical analysis of abdominal skeletal muscle and fat on computed tomography scans. Some removed adenocarcinomas were transplanted in nude mice to see their impacts on host energy homeostasis. Separately, nude mice carrying tumor grafts of MiaPaCa-2 pancreatic adenocarcinoma cells were treated with the glycolytic inhibitor 3-bromopyruvate and with emodin that inhibited glycolysis by decreasing hypoxia-inducible factor-1α. Adenocarcinomas in both group LG and group HG impaired energy homeostasis in the cancer patients, compared to the non-cancer reference. The impaired energy homeostasis induced by the adenocarcinomas in group HG was more pronounced than that by the adenocarcinomas in group LG. When original adenocarcinomas were grown in nude mice, their glycolytic abilities determined the levels of hepatic gluconeogenesis, skeletal muscle proteolysis, adipose-tissue lipolysis, and weight loss in the mice. When MiaPaCa-2 cells were grown as tumors in nude mice, 3-bromopyruvate and emodin decreased tumor-induced glycolysis and cachexia, with the best effects being seen when the drugs were administered in combination. In conclusion, the Warburg effect in pancreatobiliary adenocarcinoma triggers cancer cachexia.

Rare case of simultaneous occurrence of chronic neutrophil leukemia and T lymphoblastic lymphoma: case report and literature review.

Chronic neutrophil leukemia (CNL) is a rare and life-threatening disease. Cases of CNL combined with lymphoma are rare. Here, we report a case of CNL with T-acute lymphoblastic leukemia/lymphoma (T-ALL/LBL) in a 28-year-old male. After a regimen of ruxolitinib, VICLP (Vincristine, Idarubicin, Cyclophosphamide, Prednisone, Peg-asparaginase) regimen, high-dose cytarabine, and methotrexate regimens, the patient's bone marrow condition partially resolved. However, when the disease relapsed four months later, despite attempts with selinexor, venetoclax, and CAG(aclarubicin hydrochloride, Algocytidine, Granulocyte Stimulating Factor) chemotherapy, the leukocytes and peripheral blood primitive cells reduced, but the bone marrow did not achieve remission. This pathogenesis may be related to microenvironmental immune escape under prolonged inflammatory stimulation and gene disruption affecting protein function due to colony-stimulating factor 3 receptor gene (CSF3R) mutations. For this type of disease, early intervention may delay disease progression.

HECTD2 as a target for veratric acid in the regulation of ferroptosis in renal cell carcinoma.

Function of HECTD2 in renal cell carcinoma malignant progression is undefined. Molecular mechanism behind anti-cancer effects of veratric acid (VA) from traditional Chinese medicine (TCM) is underexplored. The Cancer Genome Atlas was leveraged to study HECTD2 expression in renal cell carcinoma and its relationship with histological grading. Kaplan-Meier survival analysis was performed. HECTD2 expression was detected in cancer cells and tissues via qRT-PCR and immunohistochemistry. GPX4 and SLC7A11 expression in tumor samples with high or low HECTD2 expression was examined by immunohistochemistry, cell viability by CCK8, cell proliferation by colony formation assay, lipid ROS and mitochondrial superoxide levels by flow cytometry, Fe2+ and MDA content by assay kits, and GPX4 and SLC7A11 proteins by western blot. SeeSAR software screened TCM small molecule compounds with highest affinity to HECTD2, confirmed with cellular thermal shift assay. VA IC50 was measured by CCK8. Xenograft model was developed and treated with VA. Tumor size and weight were monitored, with immunohistochemistry to detect HECTD2 expression in tumors and assess ferroptosis-related markers. HECTD2 was overexpressed in tumor tissues and cells, which positively correlated with histological grading. HECTD2 depletion inhibited cell vitality and proliferation, raised intracellular lipid ROS, mitochondrial superoxide, Fe2+, and MDA. HECTD2 was a target with highest VA affinity. In vitro and vivo experiments concurred that VA treatment hindered malignancy of renal cell carcinoma and enhanced its susceptibility to ferroptosis. HECTD2 supports ferroptosis resistance in renal cell carcinoma, but VA, through its targeting of HECTD2, initiates ferroptosis, showcasing its anti-cancer efficacy.

The GG genotype of rs743572 in CYP17A1 gene regulating the decrease of T/E ratio can be an independent risk factor for MetS-BPH: a retrospective cohort study.

PURPOSE: To confirm if the CYP17A1 gene regulates the ratio of T/E leading to MetS-BPH. METHODS: 824 men, aged 47-88 years, were recruited into this study through consecutive routine physical examination programs and long-term outpatient screening. Several parameters, including SNPs of CYP17A1 gene, total testosterone, estradiol, and the ratio of total testosterone to estradiol (T/E) were obtained for each participant. Based on the diagnosis of BPH, MetS, and MetS-BPH, the participants were divided into BPH and non-BPH groups, MetS and non-MetS groups, and MetS-BPH and non-MetS-BPH groups. Values of the obtained parameters were evaluated using one-way analysis of variance, Student's t-test, Chi-squared test, and logistic regression analysis. RESULTS: SNPs of the CYP17A1 gene, including the rs743572 genotypes (GG, GA, and AA), rs3781287 genotypes (GG, GT, TT), and rs4919686 genotypes (CC, CA, and AA), were present in every group. Only the GG genotype of rs743572 was independently associated with BPH (OR = 5.868, 95% CI: 3.363-7.974, P < 0.001), MetS (OR = 7.228, 95% CI: 3.925-11.331, P < 0.001), and MetS-BPH (OR = 3.417, 95% CI: 1.783-5.266, P < 0.001) after adjusting for age. In the population of genotype GG of rs743572, the decrease in T/E ratio was an independent risk factor for BPH (OR = 839.756, 95% CI: 36.978-1334.263, P = 0.001), MetS (OR = 376.988, 95% CI: 12.980-488.976, P < 0.003), and MetS-BPH (OR = 388.236, 95% CI: 24.869-495.363, P = 0.003). CONCLUSION: The GG genotype of rs743572 in CYP17A1 gene regulating the decrease of T/E ratio can be an independent risk factor for MetS-BPH populations. TRIAL REGISTRATION NUMBER: ChiCTR2200057632 "retrospectively registered". DATE OF REGISTRATION: March 15, 2022 "retrospectively registered".

O-H Insertion of Hydrogenphosphate Derivatives and α-Diazo Compounds.

An efficient O-H insertion of hydrogenphosphate derivatives and α-diazo compounds has been developed to construct α-phosphoryloxy scaffolds. Diverse α-phosphoryloxy skeletons could be obtained under mild and catalyst-free conditions in good yields. The control experiments suggest a protonation and nucleophilic addition process of α-diazo compounds via a diazonium ion pair for this transformation.

FraHMT: A Fragment-Oriented Heterogeneous Graph Molecular Generation Model for Target Proteins.

The molecular generation task stands as a pivotal step in the domains of computational chemistry and drug discovery, aiming to computationally generate molecular structures for specific properties. In contrast to previous models that focused primarily on SMILES strings or molecular graphs, our model placed a special emphasis on the substructure information on molecules, enabling the model to learn richer chemical rules and structure features from fragments and chemical reaction information on molecules. To accomplish this, we fragmented the molecules to construct heterogeneous graph representations based on atom and fragment information. Then our model mapped the heterogeneous graph data into a latent vector space by using an encoder and employed a self-regressive generative model as a decoder for molecular generation. Additionally, we performed transfer learning on the model using a small set of ligand molecules known to be active against the target protein to generate molecules that bind better to the target protein. Experimental results demonstrate that our model is highly competitive with state-of-the-art models. It can generate valid and diverse molecules with favorable physicochemical properties and drug-likeness. Importantly, they produce novel molecules with high docking scores against the target proteins.

DNMG: Deep molecular generative model by fusion of 3D information for de novo drug design.

Deep learning is improving and changing the process of de novo molecular design at a rapid pace. In recent years, great progress has been made in drug discovery and development by using deep generative models for de novo molecular design. However, most of the existing methods are string-based or graph-based and are limited by the lack of some very important properties, such as the three-dimensional information of molecules. We propose DNMG, a deep generative adversarial network (GAN) combined with transfer learning. Specifically, we use a Wasserstein-variant GAN based network architecture that considers the 3D grid spatial information of the ligand with atomic physicochemical properties to generate a representation of the molecule, which is then parsed into SMILES strings using an improved captioning network. Comprehensive in experiments demonstrate the ability of DNMG to generate valid and novel drug-like ligands. The DNMG model is used to design inhibitors for three targets, MK14, FNTA, and CDK2. The computational results show that the molecules generated by DNMG have better binding ability to the target proteins and better physicochemical properties. Overall, our deep generative model has excellent potential to generate molecules with high binding affinity for targets and explore the space of drug-like chemistry.