Artificial Intelligence in Pharmaceutical Sciences

Drug discovery and development affects various aspects of human health and dramatically impacts the pharmaceutical market.However,investments in a new drug often go unrewarded due to the long and complex process of drug research and development(R&D).With the advancement of experimental technology and computer hardware,artificial intelligence(AI)has recently emerged as a leading tool in analyzing abundant and high-dimensional data.Explosive growth in the size of biomedical data provides advantages in applying AI in all stages of drug R&D.Driven by big data in biomedicine,AI has led to a revolution in drug R&D,due to its ability to discover new drugs more efficiently and at lower cost.This review begins with a brief overview of common AI models in the field of drug discovery;then,it summarizes and discusses in depth their specific applications in various stages of drug R&D,such as target discovery,drug discovery and design,preclinical research,automated drug synthesis,and influences in the pharmaceutical market.Finally,the major limitations of AI in drug R&D are fully discussed and possible solutions are proposed.

A High-Performance Liquid Chromatography Method for the Simultaneous Determination of Five Index Components in Danhong Injection

The purpose of this study was to establish a high-performance liquid chromatography (HPLC) method for the simultaneous determination of sodium danshensu, protocatechuic aldehyde, rosmarinic acid, salvianolic acid B, and 4-coumaric acid in Danhong injection. The chromatographic method employed was as follows: the column was a Welch Ultimate XB-C18 column (250 mm × 4.6 mm, 10 μm), the mobile phase was a gradient elution of 0.4% formic acid aqueous solution (A) and acetonitrile (B), the detection wavelengths were 280 nm for sodium danshensu, protocatechuic aldehyde, and salvianolic acid B and 326 nm for 4-coumaric acid and rosmarinic acid, the sample volume was 10 μL, the flow rate was 1.0 mL/min, and the column temperature was 35°C. This method can realize the separation and determination of sodium danshensu, protocatechuic aldehyde, rosmarinic acid, salvianolic acid B, and 4-coumaric acid within 50 minutes. The linear relationships of the five peak areas and their concentrations are good (R2> 0.9997). The precision RSD values are all less than 1.0%. The reproducibility RSD values are all less than 1.3%. The stability RSD values are all less than 2.2%. The recovery values ranged from 92.4% to 99.4%. This method is simple, accurate, and reproducible. It can be used for the determination of sodium danshensu, protocatechuic aldehyde, rosmarinic acid, salvianolic acid B, and 4-coumaric acid in Danhong injection.

A Transformer-Based Ensemble Framework for the Prediction of Protein-Protein Interaction Sites

The identification of protein-protein interaction (PPI) sites is essential in the research of protein function and the discovery of new drugs. So far, a variety of computational tools based on machine learning have been developed to accelerate the identification of PPI sites. However, existing methods suffer from the low predictive accuracy or the limited scope of application. Specifically, some methods learned only global or local sequential features, leading to low predictive accuracy, while others achieved improved performance by extracting residue interactions from structures but were limited in their application scope for the serious dependence on precise structure information. There is an urgent need to develop a method that integrates comprehensive information to realize proteome-wide accurate profiling of PPI sites. Herein, a novel ensemble framework for PPI sites prediction, EnsemPPIS, was therefore proposed based on transformer and gated convolutional networks. EnsemPPIS can effectively capture not only global and local patterns but also residue interactions. Specifically, EnsemPPIS was unique in (a) extracting residue interactions from protein sequences with transformer and (b) further integrating global and local sequential features with the ensemble learning strategy. Compared with various existing methods, EnsemPPIS exhibited either superior performance or broader applicability on multiple PPI sites prediction tasks. Moreover, pattern analysis based on the interpretability of EnsemPPIS demonstrated that EnsemPPIS was fully capable of learning residue interactions within the local structure of PPI sites using only sequence information. The web server of EnsemPPIS is freely available at http://idrblab.org/ensemppis.

The impact of lipids on the cancer—immunity cycle and strategies for modulating lipid metabolism to improve cancer immunotherapy

Lipids have been found to modulate tumor biology,including proliferation,survival,and metastasis.With the new understanding of tumor immune escape that has developed in recent years,the influence of lipids on the cancer—immunity cycle has also been gradually discovered.First,regarding antigen presentation,cholesterol prevents tumor antigens from being identified by antigen presenting cells.Fatty acids reduce the expression of major histocompatibility complex class I and costimulatory factors in dendritic cells,impairing antigen presentation to T cells.Prostaglandin E2(PGE2)reduce the accumulation of tumor-infiltrating dendritic cells.Regarding T-cell priming and activation,cholesterol destroys the structure of the T-cell receptor and reduces immunodetection.In contrast,cholesterol also promotes T-cell receptor clustering and relative signal transduction.PGE2 represses T-cell proliferation.Finally,regarding T-cell killing of cancer cells,PGE2 and cholesterol weaken granule-dependent cytotoxicity.Moreover,fatty acids,cholesterol,and PGE2 can improve the activity of immunosuppressive cells,increase the expression of immune checkpoints and promote the secretion of immunosuppressive cytokines.Given the regulatory role of lipids in the cancer—immunity cycle,drugs that modulate fatty acids,cholesterol and PGE2 have been envisioned as effective way in restoring antitumor immunity and synergizing with immunotherapy.These strategies have been studied in both preclinical and clinical studies.

MF-SuP-pK_(a): Multi-fidelity modeling with subgraph pooling mechanism for pK_(a) prediction

Acid-base dissociation constant(pK_(a)) is a key physicochemical parameter in chemical science, especially in organic synthesis and drug discovery. Current methodologies for pK_(a) prediction still suffer from limited applicability domain and lack of chemical insight. Here we present MF-SuP-pK_(a)(multi-fidelity modeling with subgraph pooling for pK_(a) prediction), a novel pK_(a) prediction model that utilizes subgraph pooling, multi-fidelity learning and data augmentation. In our model, a knowledgeaware subgraph pooling strategy was designed to capture the local and global environments around the ionization sites for micro-pK_(a) prediction. To overcome the scarcity of accurate pK_(a) data, lowfidelity data(computational pK_(a)) was used to fit the high-fidelity data(experimental pK_(a)) through transfer learning. The final MF-SuP-pK_(a) model was constructed by pre-training on the augmented ChEMBL data set and fine-tuning on the DataWarrior data set. Extensive evaluation on the DataWarrior data set and three benchmark data sets shows that MF-SuP-pK_(a) achieves superior performances to the state-of-theart pK_(a) prediction models while requires much less high-fidelity training data. Compared with Attentive FP, MF-SuP-pK_(a) achieves 23.83% and 20.12% improvement in terms of mean absolute error(MAE) on the acidic and basic sets, respectively.

Data-driven structural descriptor for predicting platinum-based alloys as oxygen reduction electrocatalysts

Owing to increasing global demand for carbon neutral and fossil-free energy systems,extensive research is being conducted on efficient and inexpensive electrocatalysts for catalyzing the kinetically sluggish oxygen reduction reaction(ORR)at the cathode of fuel cells.Platinum(Pt)-based alloys are considered promising candidates for replacing expensive Pt catalysts.However,the current screening process of Pt-based alloys is time-consuming and labor-intensive,and the descriptor for predicting the activity of Pt-based catalysts is generally inaccurate.This study proposed a strategy by combining high-throughput first-principles calculations and machine learning to explore the descriptor used for screening Pt-based alloy catalysts with high Pt utilization and low Pt consump-tion.Among the 77 prescreened candidates,we identified 5 potential candidates for catalyzing ORR with low overpotential.Furthermore,during the second and third rounds of active learning,more Pt-based alloys ORR candidates are identi-fied based on the relationship between structural features of Pt-based alloys and their activity.In addition,we highlighted the role of structural features in Pt-based alloys and found that the difference between the electronegativity of Pt and heteroatom,the valence electrons number of the heteroatom,and the ratio of heteroatoms around Pt are the main factors that affect the activity of ORR.More importantly,the combination of those structural features can be used as structural descriptor for predicting the activity of Pt-based alloys.We believe the findings of this study will provide new insight for predicting ORR activ-ity and contribute to exploring Pt-based electrocatalysts with high Pt utiliza-tion and low Pt consumption experimentally.

Deubiquitination complex platform:A plausible mechanism for regulating the substrate specificity of deubiquitinating enzymes

Deubiquitinating enzymes(DUBs) or deubiquitinases facilitate the escape of multiple proteins from ubiquitin-proteasome degradation and are critical for regulating protein expression levels in vivo.Therefore,dissecting the underlying mechanism of DUB recognition is needed to advance the development of drugs related to DUB signaling pathways.To data,extensive studies on the ubiquitin chain specificity of DUBs have been reported,but substrate protein recognition is still not clearly understood.As a breakthrough,the scaffolding role may be significant to substrate protein selectivity.From this perspective,we systematically characterized the scaffolding proteins and complexes contributing to DUB substrate selectivity.Furthermore,we proposed a deubiquitination complex platform(DCP) as a potentially generic mechanism for DUB substrate recognition based on known examples,which might fill the gaps in the understanding of DUB substrate specificity.

Targeted genome mining for microbial antitumor agents acting through DNA intercalation

Microbial natural products have been one of the most important sources for drug development.In the current postgenomic era,sequence-driven approaches for natural product discovery are becoming increasingly popular.Here,we develop an effective genome mining strategy for the targeted discovery of microbial metabolites with antitumor activities.Our method employs uvrA-like genes as genetic markers,which have been identified in the biosynthetic gene clusters(BGCs)of several chemotherapeutic drugs of microbial origin and confer self-resistance to the corresponding producers.Through systematic genomic analysis of gifted actinobacteria genera,identification of uvrA-like gene-containing BGCs,and targeted isolation of products from a BGC prioritized for metabolic analysis,we identified a new tetracycline-type DNA intercalator timmycins.Our results thus provide a new genome mining strategy for the efficient discovery of antitumor agents acting through DNA intercalation.

Microsnoop: A generalist tool for microscopy image representation

Accurate profiling of microscopy images from small scale to high throughput is an essential procedure in basic and applied biological research.Here,we present Microsnoop,a novel deep learning–based representation tool trained on large-scale microscopy images using masked self-supervised learning.Microsnoop can process various complex and heterogeneous images,and we classified images into three categories:single-cell,full-field,and batch-experiment images.Our benchmark study on 10 high-quality evaluation datasets,containing over 2,230,000 images,demonstrated Microsnoop’s robust and state-ofthe-art microscopy image representation ability,surpassing existing generalist and even several custom algorithms.Microsnoop can be integrated with other pipelines to perform tasks such as superresolution histopathology image and multimodal analysis.Furthermore,Microsnoop can be adapted to various hardware and can be easily deployed on local or cloud computing platforms.We will regularly retrain and reevaluate the model using communitycontributed data to consistently improve Microsnoop.

Pushing the Boundaries of Molecular Property Prediction for Drug Discovery with Multitask Learning BERT Enhanced by SMILES Enumeration

Accurate prediction of pharmacological properties of small molecules is becoming increasingly important in drug discovery.Traditional feature-engineering approaches heavily rely on handcrafted descriptors and/or fingerprints,which need extensive human expert knowledge.With the rapid progress of artificial intelligence technology,data-driven deep learning methods have shown unparalleled advantages over feature-engineering-based methods.However,existing deep learning methods usually suffer from the scarcity of labeled data and the inability to share information between different tasks when applied to predicting molecular properties,thus resulting in poor generalization capability.Here,we proposed a novel multitask learning BERT(Bidirectional Encoder Representations from Transformer)framework,named MTL-BERT,which leverages large-scale pre-training,multitask learning,and SMILES(simplified molecular input line entry specification)enumeration to alleviate the data scarcity problem.MTL-BERT first exploits a large amount of unlabeled data through self-supervised pretraining to mine the rich contextual information in SMILES strings and then fine-tunes the pretrained model for multiple downstream tasks simultaneously by leveraging their shared information.Meanwhile,SMILES enumeration is used as a data enhancement strategy during the pretraining,fine-tuning,and test phases to substantially increase data diversity and help to learn the key relevant patterns from complex SMILES strings.The experimental results showed that the pretrained MTL-BERT model with few additional fine-tuning can achieve much better performance than the state-of-the-art methods on most of the 60 practical molecular datasets.Additionally,the MTL-BERT model leverages attention mechanisms to focus on SMILES character features essential to target properties for model interpretability.

The inhibitory effect of adenosine on tumor adaptive immunity and intervention strategies

Adenosine(Ado)is significantly elevated in the tumor microenvironment(TME)compared to normal tissues.It binds to adenosine receptors(AdoRs),suppressing tumor antigen presentation and immune cell activation,thereby inhibiting tumor adaptive immunity.Ado downregulates major histocompatibility complex II(MHC II)and co-stimulatory factors on dendritic cells(DCs)and macrophages,inhibiting antigen presentation.It suppresses anti-tumor cytokine secretion and T cell activation by disrupting T cell receptor(TCR)binding and signal transduction.Ado also inhibits chemokine secretion and KCa3.1 channel activity,impeding effector T cell trafficking and infiltration into the tumor site.Furthermore,Ado diminishes T cell cytotoxicity against tumor cells by promoting immune-suppressive cytokine secretion,upregulating immune checkpoint proteins,and enhancing immune-suppressive cell activity.Reducing Ado production in the TME can significantly enhance anti-tumor immune responses and improve the efficacy of other immunotherapies.Preclinical and clinical development of inhibitors targeting Ado generation or AdoRs is underway.Therefore,this article will summarize and analyze the inhibitory effects and molecular mechanisms of Ado on tumor adaptive immunity,as well as provide an overview of the latest advancements in targeting Ado pathways in anti-tumor immune responses.

Deciphering the Shared and Specific Drug Resistance Mechanisms of Anaplastic Lymphoma Kinase via Binding Free Energy Computation

Anaplastic lymphoma kinase(ALK),a tyrosine receptor kinase,has been proven to be associated with the occurrence of numerous malignancies.Although there have been already at least 3 generations of ALK inhibitors approved by FDA or in clinical trials,the occurrence of various mutations seriously attenuates the effectiveness of the drugs.Unfortunately,most of the drug resistance mechanisms still remain obscure.Therefore,it is necessary to reveal the bottom reasons of the drug resistance mechanisms caused by the mutations.In this work,on the basis of verifying the accuracy of 2 main kinds of binding free energy calculation methodologies[end-point method of Molecular Mechanics with Poisson-Boltzmann/Generalized Born and Surface Area(MM/PB(GB)SA)and alchemical method of Thermodynamic Integration(TI)],we performed a systematic analysis on the ALK systems to explore the underlying shared and specific drug resistance mechanisms,covering the one-drug-multiple-mutation and multiple-drug-onemutation cases.

Enhancing Generic Reaction Yield Prediction through Reaction Condition-Based Contrastive Learning

Deep learning(DL)-driven efficient synthesis planning may profoundly transform the paradigm for designing novel pharmaceuticals and materials.However,the progress of many DL-assisted synthesis planning(DASP)algorithms has suffered from the lack of reliable automated pathway evaluation tools.As a critical metric for evaluating chemical reactions,accurate prediction of reaction yields helps improve the practicality of DASP algorithms in the real-world scenarios.Currently,accurately predicting yields of interesting reactions still faces numerous challenges,mainly including the absence of high-quality generic reaction yield datasets and robust generic yield predictors.To compensate for the limitations of high-throughput yield datasets,we curated a generic reaction yield dataset containing 12 reaction categories and rich reaction condition information.Subsequently,by utilizing 2 pretraining tasks based on chemical reaction masked language modeling and contrastive learning,we proposed a powerful bidirectional encoder representations from transformers(BERT)-based reaction yield predictor named Egret.It achieved comparable or even superior performance to the best previous models on 4 benchmark datasets and established state-of-the-art performance on the newly curated dataset.We found that reaction-condition-based contrastive learning enhances the model’s sensitivity to reaction conditions,and Egret is capable of capturing subtle differences between reactions involving identical reactants and products but different reaction conditions.Furthermore,we proposed a new scoring function that incorporated Egret into the evaluation of multistep synthesis routes.Test results showed that yield-incorporated scoring facilitated the prioritization of literature-supported high-yield reaction pathways for target molecules.In addition,through meta-learning strategy,we further improved the reliability of the model’s prediction for reaction types with limited data and lower data quality.Our results suggest that Egret holds the potential to become an essential component of the next-generation DASP tools.

Josephin domain containing 2(JOSD2)promotes lung cancer by inhibiting LKB1(Liver kinase B1)activity

Non-small cell lung cancer(NSCLC)ranks as one of the leading causes of cancer-related deaths worldwide.Despite the prominence and effectiveness of kinase-target therapies in NSCLC treatment,these drugs are suitable for and beneficial to a mere~30%of NSCLC patients.Consequently,the need for novel strategies addressing NSCLC remains pressing.Deubiquitinases(DUBs),a group of diverse enzymes with well-defined catalytic sites that are frequently overactivated in cancers and associated with tumorigenesis and regarded as promising therapeutic targets.Nevertheless,the mechanisms by which DUBs promote NSCLC remain poorly understood.Through a global analysis of the 97 DUBs’contribution to NSCLC survival possibilities using The Cancer Genome Atlas(TCGA)database,we found that high expression of Josephin Domain-containing protein 2(JOSD2)predicted the poor prognosis of patients.Depletion of JOSD2 significantly impeded NSCLC growth in both cell/patient-derived xenografts in vivo.Mechanically,we found that JOSD2 restricts the kinase activity of LKB1,an important tumor suppressor generally inactivated in NSCLC,by removing K6-linked polyubiquitination,an action vital for maintaining the integrity of the LKB1-STRAD-MO25 complex.Notably,we identified the first small-molecule inhibitor of JOSD2,and observed that its pharmacological inhibition significantly arrested NSCLC proliferation in vitro/in vivo.Our findings highlight the vital role of JOSD2 in hindering LKB1 activity,underscoring the therapeutic potential of targeting JOSD2 in NSCLC,especially in those with inactivated LKB1,and presenting its inhibitors as a promising strategy for NSCLC treatment.

Generic Interpretable Reaction Condition Predictions with Open Reaction Condition Datasets and Unsupervised Learning of Reaction Center

Effective synthesis planning powered by deep learning(DL)can significantly accelerate the discovery of new drugs and materials.However,most DL-assisted synthesis planning methods offer either none or very limited capability to recommend suitable reaction conditions(RCs)for their reaction predictions.Currently,the prediction of RCs with a DL framework is hindered by several factors,including:(a)lack of a standardized dataset for benchmarking,(b)lack of a general prediction model with powerful representation,and(c)lack of interpretability.To address these issues,we first created 2 standardized RC datasets covering a broad range of reaction classes and then proposed a powerful and interpretable Transformer-based RC predictor named Parrot.Through careful design of the model architecture,pretraining method,and training strategy,Parrot improved the overall top-3 prediction accuracy on catalysis,solvents,and other reagents by as much as 13.44%,compared to the best previous model on a newly curated dataset.Additionally,the mean absolute error of the predicted temperatures was reduced by about 4℃.Furthermore,Parrot manifests strong generalization capacity with superior cross-chemical-space prediction accuracy.Attention analysis indicates that Parrot effectively captures crucial chemical information and exhibits a high level of interpretability in the prediction of RCs.The proposed model Parrot exemplifies how modern neural network architecture when appropriately pretrained can be versatile in making reliable,generalizable,and interpretable recommendation for RCs even when the underlying training dataset may still be limited in diversity.

The dichotomous role of immunoproteasome in cancer:Friend or foe?

Immunoproteasome is a variant of proteasome with structural differences in 20S subunits optimizing them for the production of antigenic peptides with higher binding affinity to major histocompatibility complex(MHC)-I molecules.Apart from this primary function in antigen presentation,immunoproteasome is also responsible for the degradation of proteins,both unfolded proteins for the maintenance of protein homeostasis and tumor suppressor proteins contributing to tumor progression.The altered expression of immunoproteasome is frequently observed in cancers;however,its expression levels and effects vary among different cancer types exhibiting antagonistic roles in tumor development.This review focuses on the dichotomous role of immunoproteasome in different cancer types,as well as summarizes the current progression in immunoproteasome activators and inhibitors.Specifically targeting immunoproteasome may be a beneficial therapeutic intervention in cancer treatment and understanding the role of immunoproteasome in cancers will provide a significant therapeutic insight for the prevention and treatment of cancers.

Kinome-wide polypharmacology profiling of small molecules by multi-task graph isomorphism network approach

Prediction of the interactions between small molecules and their targets play important roles in various applications of drug development,such as lead discovery,drug repurposing and elucidation of potential drug side effects.Therefore,a variety of machine learning-based models have been developed to predict these interactions.In this study,a model called auxiliary multi-task graph isomorphism network with uncertainty weighting(AMGU)was developed to predict the inhibitory activities of small molecules against 204 different kinases based on the multi-task Graph Isomorphism Network(MT-GIN)with the auxiliary learning and uncertainty weighting strategy.The calculation results illustrate that the AMGU model outperformed the descriptor-based models and state-of-the-art graph neural networks(GNN)models on the internal test set.Furthermore,it also exhibited much better performance on two external test sets,suggesting that the AMGU model has enhanced generalizability due to its great transfer learning capacity.Then,a naÏve model-agnostic interpretable method for GNN called edges masking was devised to explain the underlying predictive mechanisms,and the consistency of the interpretability results for 5typical epidermal growth factor receptor(EGFR)inhibitors with their structure-activity relationships could be observed.Finally,a free online web server called KIP was developed to predict the kinomewide polypharmacology effects of small molecules(http://cadd.zju.edu.cn/kip).

WSB1 regulates c-Myc expression through β-catenin signaling and forms a feedforward circuit

The dysregulation of transcription factors is widely associated with tumorigenesis.As the most well-defined transcription factor in multiple types of cancer,c-Myc can transform cells by transactivating various downstream genes.Given that there is no effective way to directly inhibit c-Myc,c-Myc targeting strategies hold great potential for cancer therapy.In this study,we found that WSB1,which has a highly positive correlation with c-Myc in 10 cancer cell lines and clinical samples,is a direct target gene of c-Myc,and can positively regulate c-Myc expression,which forms a feedforward circuit promoting cancer development.RNA sequencing results from Bel-7402 cells confirmed that WSB1 promoted cMyc expression through theβ-catenin pathway.Mechanistically,WSB1 affectedβ-catenin destruction complex-PPP2CA assembly and E3 ubiquitin ligase adaptorβ-TRCP recruitment,which inhibited the ubiquitination ofβ-catenin and transactivated c-Myc.Of interest,the effect of WSB1 on c-Myc was independent of its E3 ligase activity.Moreover,overexpressing WSB1 in the Bel-7402 xenograft model could further strengthen the tumor-driven effect of c-Myc overexpression.Thus,our findings revealed a novel mechanism involved in tumorigenesis in which the WSB1/c-Myc feedforward circuit played an essential role,highlighting a potential c-Myc intervention strategy in cancer treatment.

Disulfiram enhances the antitumor activity of cisplatin by inhibiting the Fanconi anemia repair pathway

A series of chemotherapeutic drugs that induce DNA damage,such as cisplatin(DDP),are standard clinical treatments for ovarian cancer,testicular cancer,and other diseases that lack effective targeted drug therapy.Drug resistance is one of the main factors limiting their application.Sensitizers can overcome the drug resistance of tumor cells,thereby enhancing the antitumor activity of chemotherapeutic drugs.In this study,we aimed to identify marketable drugs that could be potential chemotherapy sensitizers and explore the underlying mechanisms.We found that the alcohol withdrawal drug disulfiram(DSF)could significantly enhance the antitumor activity of DDP.JC-1 staining,propidium iodide(PI)staining,and western blotting confirmed that the combination of DSF and DDP could enhance the apoptosis of tumor cells.Subsequent RNA sequencing combined with Gene Set Enrichment Analysis(GSEA)pathway enrichment analysis and cell biology studies such as immunofluorescence suggested an underlying mechanism:DSF makes cells more vulnerable to DNA damage by inhibiting the Fanconi anemia(FA)repair pathway,exerting a sensitizing effect to DNA damaging agents including platinum chemotherapy drugs.Thus,our study illustrated the potential mechanism of action of DSF in enhancing the antitumor effect of DDP.This might provide an effective and safe solution for combating DDP resistance in clinical treatment.

Author correction to‘Mevalonate improves anti-PD-1/PD-L1 efficacy by stabilizing CD274 mRNA'[Acta Pharmaceutica Sinica B 13(2023)2585-2600]

The authors regret that one author was missed in the author list when preparing the manuscript,while the foundation to him was mentioned in the section of Acknowledgments.Dr.Qinjie Weng should be added to the author list because of his contribution in animal study and his foundation help us to complete this study.The authors sincerely apologize for any inconvenience caused to the journal and readers.