检索结果-南通市图书馆

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

引用

Research 2023年第3期2022卷 157-170页

作者： Xiao-Chen Zhang Cheng-Kun Wu Jia-Cai Yi Xiang-Xiang Zeng Can-Qun Yang Ai-Ping Lu Ting-Jun Hou dong-sheng cao Xiangya School of Pharmaceutical Sciences Central South UniversityChangsha 410013Hunan P.R.China Shangqiu Normal University School of Information TechnologyShangqiu 476000HenanP.R.China College of Computer National University of Defense TechnologyChangsha 410005HunanP.R.China Department of Computer Science Hunan UniversityChangsha 410082.HunanP.R.China institute for Advancing Translational Medicine in Bone and Joint Diseases School of Chinese MedicineHong Kong Baptist UniversityHong Kong SAR 999077P.R.China Innovation Institute for Artificial Intelligence in Medicine of Zhejiang University.College of Pharmaceutical Sciences Zhejiang UniversityHangzhou 310058ZhejiangP.R.China

Accurate prediction of pharmacological properties of small molecules is becoming increasingly important in drug *** feature-engineering approaches heavily rely on handcrafted descriptors and/or fingerprints,which need extensive human expert *** the rapid progress of artificial intelligence technology,data-driven deep learning methods have shown unparalleled advantages over feature-engineering-based ***,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 ***,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 ***-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 ***,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 *** 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 ***,the MTL-BERT model leverages attention mechanisms to focus on SMILES character features essential to target properties for model interpretability.

关键词： simplified generalization tuning

来源：

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Systematic Evaluation of logD7.4 based on Ensemble Machine Learning, Group Contribution and Matched Molecular Pair Analysis

Systematic Evaluation of logD7.4 based on Ensemble Machine L...

引用

2019中国化学会第十五届全国计算（机）化学学术会议

作者：付丽 dong-sheng cao 中南大学 Xiangya School of Pharmaceutical Sciences Central South University

Lipophilicity,as evaluated by the n-octanol/buffer solution distribution coefficient at pH=7.4(logD),is a major determinant of various absorption,distribution,metabolism,elimination and toxicology (ADMET) parameters of drug *** this study,we developed several quantitative structure-property relationship (QSPR) models to predict logD based on a large and structurally diverse data *** popular machine learning algorithms were employed to build the prediction models with 43 molecular descriptors selected by a wrapper feature selection *** results demonstrated XGBoost yielded better prediction performance than any other single model (R=0.912 and RMSE=0.382).However,the consensus model from the top three models could continue to improve the prediction performance (R=0.923 and RMSE=0.359).The robustness,reliability,and generalization ability of the models were strictly evaluated by the Y-randomization test and applicability domain ***,the group contribution model based on 110 atom types and the local models for different ionization states were also established and compared with the global *** results demonstrated that the descriptor-based consensus model is superior to the group contribution method,and the local models have no advantage over the global ***,matched molecular pair (MMP) analysis and descriptor importance analysis were performed to extract transformation rules and give some explanations related to *** conclusion,we believe that the consensus model developed in this study can be used as a reliable and promising tool to evaluate logD in drug discovery.

关键词： logD7.4 XGBoost Group Contribution MMP Ensemble Learning

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

Three-Level Hepatotoxicity Prediction System Based on Adverse Hepatic Effects

Three-Level Hepatotoxicity Prediction System Based on Advers...

引用

2019中国化学会第十五届全国计算（机）化学学术会议

作者：刘璐 dong-sheng cao 中南大学 Xiangya School of Pharmaceutical Sciences Central South University

Hepatotoxicity is a major cause of drug withdrawal from the market. To reduce the drug attrition induced by hepatotoxicity, an accurate and efficient hepatotoxicity prediction system must be constructed. In the present study, we constructed a three-level hepatotoxicity prediction system based on different levels of adverse hepatic effects（AHEs） combined with machine learning, using: 1, an endpoint: hepatotoxicity;2, four hepatotoxicity severity degrees and 3, specific AHEs. After collecting and curing 15,873 compound-AHE pairs associated with 2017 compounds and 403 AHEs, we constructed 27 models with three endpoint levels with the random forest algorithm, and obtained accuracies ranging from 67.0% to 78.2% and the area under receiver operating characteristic curves（AUCs） of 0.715-0.875. The 27 models were fully integrated into a tiered hepatotoxicity prediction system. The existence of hepatotoxicity existence, severity degree and potential AHEs for a given compound could be inferred simultaneously and systematically. Thus, the tiered hepatotoxicity prediction system allows researchers to have significant confidence in confirming compound hepatotoxicity, analyze hepatotoxicity from multiple perspectives, obtain warnings for the potential hepatotoxicity severity and even rapidly select the proper in vitro experiments for hepatotoxicity verification. We also applied three external sets（11 drugs or candidates that failed in clinical trials or were withdrawn from the market, the PharmGKB（offsides） database and an herbal hepatotoxicity dataset） to test and validate the prediction ability of our system. Furthermore, the hepatotoxicity prediction system was adapted into a flow framework based on Konstanz Information Miner（KNIME）, which was made available for researchers.

关键词： hepatotoxicity adverse hepatic effects(AHEs) toxicity risk assessment SAR

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

EAT-Score: A New Strategy to Improve Docking-based Virtual Screening by Integrating Multiple Energy Auxiliary Terms from MOE

EAT-Score: A New Strategy to Improve Docking-based Virtual S...

引用

2019中国化学会第十五届全国计算（机）化学学术会议

作者：叶文玲 dong-sheng cao 中南大学 Xiangya School of Pharmaceutical Sciences Central South University

Virtual screening（VS） is an efficient and fast method for discovering novel hit compounds in drug discovery, especially the docking-based VS. However, the accuracy of current docking scoring function（SF） is usually insufficient. Therefore, combined with e Xtreme Gradient Boosting（XGBoost）, we proposed a novel approach named EAT-Score to improve the screening power of SF by directly utilizing the Energy Auxiliary Terms（EAT） of docking programs. In this study, the EAT was defined as the docking output of MOE, including the energy scores of five different classical SFs and the protein-ligand interaction fingerprint（PLIF） terms. The performance of EAT-Score to discriminate actives from decoys was strictly validated on DUD-E diverse subset by using different performance metrics. The results showed that EAT-Score performed much better than traditional classical SFs in VS, with its AUC values exhibiting around a 0.3 improvement. Meanwhile, EAT-Score could achieve substantial improvements compared to other state-of-the-art VS methods, such as other machine learning（ML）-based SFs and docking programs, whichever in terms of AUC, EF, LogAUC or BEDROC. Furthermore, our developed EAT-Score model can capture important binding pattern information from protein-ligand complexes, which may be very helpful in binding mechanism interpretation of a certain target and thereby helps guide the further drug design.

关键词： screening power VS docking XGBoost

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

Recent advances of variable selection methods in chemical modeling

Recent advances of variable selection methods in chemical mo...

引用

2019中国化学会第十五届全国计算（机）化学学术会议

作者：云永欢 Hong-dong Li Bai-Chuan Deng dong-sheng cao 海南大学 School of Information Science and Engineering Central South University College of Animal Science South China Agricultural University Xiangya School of Pharmaceutical Sciences Central South University

With increasing development in innovative instrumentation, the chemical data is usually in the shape of high dimension. Variable（feature, wavelength） selection is thus a critical step in multivariate calibration of chemical modeling, which can improve the prediction performance, make the calibration reliable and provide simpler interpretation. During the several decades, there have been a large number of variable selection methods proposed in chemometrics, in order to address all kinds of analytical issues. In this research [1], we generalize variable selection methods in a simple manner to introduce their classifications, merits and drawbacks, to provide a better understanding of their characteristics, similarities and differences. Figure 1 shows the classification of variable selection methods in near infrared spectroscopy. We also introduce some hybrid and modified methods, highlighting their improvements. Finally, we summarize the limitations of existing variable selection methods, providing our remarks and suggestions on the development of variable selection methods.

关键词： variable selection chemical modeling chemometrics near infrared spectroscopy

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

ChemDes:a free web-based tool for computing molecular descriptors

ChemDes:a free web-based tool for computing molecular descri...

引用

第十二届全国青年药学工作者最新科研成果交流会

作者： Jie dong Shao Liu Xiong-Bo Xu Wenbin Zeng dong-sheng cao Xiangya Hospital Central South University College of Pharmaceutical Sciences Central South University

Summary:The calculation of molecular descriptors is the basic study for characterizing small molecules and is widely used in drug discovery such as QSAR/SAR and drug ADME/T *** order to provide a convenient way for computing,and facilitate extensive studies of drug molecules,we developed a web-based tool called *** allows users to compute more than 1,144 2D/3D molecular descriptors that are divided into 18 logical blocks and seven types of molecular fingerprint systems for drug molecules,including topological fingerprints,electro-topological state(E-state) fingerprints,MACCS keys,FP4 keys,atom pairs fingerprints,topological torsion fingerprints and Morgan/circular *** addition,ChemDes contains three useful auxiliary tools to achieve corresponding functions(format converting,MOP AC optimization and fingerprint similarity).

关键词：

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

基于配体和基于受体的虚拟筛选方法在PARP1抑制剂上的活性富集和骨架富集的系统性评估（英文）

基于配体和基于受体的虚拟筛选方法在PARP1抑制剂上的活性富集和...

引用

2019中国化学会第十五届全国计算（机）化学学术会议

作者：赵越 dong-sheng cao 中南大学 Xiangya School of Pharmaceutical Sciences Central South University

Poly（ADP-ribose） Polymerase1（PARP1） is one of the most investigated enzymes in 18 membered PARP families and is involved in a variety of cellular functions including DNA damage repair, gene transcription and cell apoptosis. PARP1 can form a PARP1（ADP-ribose） polymer, then bind to the DNA damage gap to recruit DNA repair proteins, and repair the break to maintain genomic stability. PARP1 is highly expressed in tumor cells, so the inhibition of PARP1 can block DNA repair, promote tumor cell apoptosis, and exert antitumor activity. To date, four PARP1 inhibitors namely olaparib, rucaparib, niraparib, talazoparib, have been approved by FDA for treating ovarian cancer and breast cancer with BRCA1/2 mutation. These drugs have showed super advantages over conventional chemotherapeutic drugs with low hematological toxicity and slowly developed drug resistance. In this article, it will highlight the majority of PARP1 inhibitors that have entered clinical trials based on parent chemical structures of cyclic lactams and exocyclic amides firstly, then collect bioactivity data of existed PARP1 inhibitors. With the matched molecular pair analysis, it will explore structure-activity relationship of newly developed PARP1 inhibitors. Therefore, the primary purpose of this study is to present information that has been found, and to promote the process in further findings regarding more potent and safer PARP1 inhibitors.

关键词： PARP1 Cancer PARP1 inhibitors Structure-Activity relationship Matched molecular pair analysis

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

BioTriangle:a web service for computing various descriptors of small molecules,proteins,DNAs/RNAs and their interactions

BioTriangle:a web service for computing various descriptors ...

引用

中国化学会第30届学术年会-第二十五分会：化学信息学与化学计量学

作者： Jie dong Zhi-Jiang Yao Ming Wen Min-Feng Zhu Ning-Ning Wang Hong-Yu Miao dong-sheng cao School of Pharmaceutical Sciences Central South University College of Chemistry and Chemical Engineering Central South University School of Mathematics and Statistics Central South University Department of Biostatistics School of Public HealthUniversity of Texas Health Science Center at Houston

Rapid advances in network biology indicate that most cellular components exert their functions through interactions with other cellular components,such as proteins,DNA,RNA and small *** rapidly increasing amount of publicly available data in biology and chemistry enables researchers to revisit interaction problems by systematic integration and analysis of heterogeneous ***,we developed a comprehensive molecular representation tool,called BioTriangle,to emphasize the integration of chemoinformatics and bioinformatics into a molecular informatics platform for computational biology *** is a feature-rich toolkit used for the characterization of various complex biological molecules and interaction samples,such as chemicals,proteins,DNA/RNA,and their *** calculates nine types of features including chemical descriptors or molecular fingerprints,structural and physicochemical features of proteins and peptides from amino acid sequence,composition and physicochemical features of DNA/RNA from their primary sequences,chemical-chemical interaction features,chemical-protein interaction features,chemical-DNA/RNA interaction features,protein-protein interaction features,protein-DNA/RNA interaction features,and DNA/RNA-DNA/RNA interaction *** recommend BioTriangle to analyze and represent various complex molecular data under *** hope that the web service will be helpful when exploring questions concerning structures,functions and interactions of various molecular data in the context of computational biology.

关键词： Molecular descriptors molecular representation interaction features QSAR/QSPR chemoinformatics

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

MXene Hybridized Polymer with Enhanced Electromagnetic Energy Harvest for Sensitized Microwave Actuation and Self‑Powered Motion Sensing

引用

Nano-Micro Letters 2025年第3期17卷 265-279页

作者： Yu‑Ze Wang Yu‑Chang Wang Ting‑Ting Liu Quan‑Liang Zhao Chen‑Sha Li Mao‑sheng cao School of Materials Science and Engineering Beijing Institute of TechnologyBeijing 100081People’s Republic of China School of Materials Science and Engineering Peking UniversityBeijing 100871People’s Republic of China School of Mechanical and Material Engineering North China University of TechnologyBeijing 100144People’s Republic of China Key Laboratory of Functional Inorganic Material Chemistry Ministry of Education of the People’s Republic of ChinaHeilongjiang UniversityHarbin 150080People’s Republic of China

Polymeric microwave actuators combining tissue-like softness with programmablemicrowave-responsive deformation hold great promise for mobile intelligentdevices and bionic soft robots. However, their application is challenged by restricted electromagneticsensitivity and intricate sensing coupling. In this study, a sensitized polymericmicrowave actuator is fabricated by hybridizing a liquid crystal polymer with Ti3C2Tx(MXene). Compared to the initial counterpart, the hybrid polymer exhibits unique spacechargepolarization and interfacial polarization, resulting in significant improvements of230% in the dielectric loss factor and 830% in the apparent efficiency of electromagneticenergy harvest. The sensitized microwave actuation demonstrates as the shortenedresponse time of nearly 10 s, which is merely 13% of that for the initial shape memory polymer. Moreover, the ultra-low content of MXene (upto 0.15 wt%) benefits for maintaining the actuation potential of the hybrid polymer. An innovative self-powered sensing prototype that combinesdriving and piezoelectric polymers is developed, which generates real-time electric potential feedback (open-circuit potential of ~ 3 mV) duringactuation. The polarization-dominant energy conversion mechanism observed in the MXene-polymer hybrid structure furnishes a new approachfor developing efficient electromagnetic dissipative structures and shows potential for advancing polymeric electromagnetic intelligent devices.

关键词： Microwave absorption Electromagnetic response Energy harvest Self-sensing Soft actuator

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Positive effect of Mo on mechanical properties of a Ni-42W-10Co-x Mo medium heavy alloy

引用

China Foundry 2025年

作者： Guo-xin cao Jian-xin dong Yong-sheng He He Jiang Fang Hao Zhi-hua Nie Teng-fei Ma Bao-quan Fu School of Materials Science and Engineering University of Science and Technology Beijing Xi'an Superalloy Technology Co. Ltd. School of Materials Science and Engineering Beijing Institute of Technology Key Laboratory of Air-driven Equipment Technology of Zhejiang Province Quzhou University

In this study,a novel Ni-W-Co-Mo medium heavy alloy (MHA) was designed to improve its mechanical strength via Mo *** the Ni-42W-10Co-xM o alloy series,where x represents the weight percent of Mo and varies between 0,1,2,5,and 10,the microstructure transitions from a dendritic structure to a hypoeutectic structure as the Mo content increases from 0 to 5wt.%.Moreover,as the Mo content increases from 0to 10wt.%,the distribution of theμ-phase shifts from being individually dispersed to forming aggregates,and its volume fraction rises from 0.5%to 7.9%.Notably,theμ-phase evolves into an eutectic microstructure,which helps in minimizing the segregation of *** change is accompanied by a substantial enhancement in mechanical properties;specifically,the compressive yield strength at room temperature increases from 350 MPa to 646 MPa,indicating a significant 85%***,the microhardness increases from 230 HV to 304 *** dynamics simulations further reveal that the strengthening mechanism of Ni-42W-10Co-xM o alloys is Mo-induced solid solution strengthening and precipitation strengthening.

关键词：

来源：

同方期刊数据库评论

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

欢迎您,

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

请选择保存的检索档案：

请选择收藏分类：

通借通还

欢迎您,

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：