中药化学数据库的新版本、数据挖掘及GABA_A受体的分子模型化
文献类型:学位论文
作者 | 陆爱军 |
学位类别 | 博士 |
答辩日期 | 2004 |
授予单位 | 中国科学院过程工程研究所 |
授予地点 | 中国科学院过程工程研究所 |
导师 | 周家驹 |
关键词 | 中药 数据库 数据挖掘 关联规则 定量药效关系 柔性原子受体模型 |
其他题名 | Updating and Mining in Traditional Chinese Medicine Chemical Database and Molecular Modeling of GABAa Receptor |
中文摘要 | 本文将数据挖掘技术引入中药及化学领域,对中药化学数据库进行了新版本的研制及数据挖掘研究,并对咪哇苯并二氮杂卓类化合物对五种重组GABAA受体亚型ax助2(x=1,2,3,5,6)的亲和力进行了分子模型化的研究。主要研究内容和结果如下:进行了数据库新版本的研制,其中包括规范集、结构框架、结构信息、中药命名系统、合法性检验等方面,给出了很多数据库的检索实例,探讨了中药化学数据库的功能和用途。统计了中药化学数据库的“类药性”性质,结果表明数据库中化合物的分子量和LogP值大部分都满足的Lipinski原则。建立了中药化学数据库的数据仓库和数据集市,对源数据进行了数据清理和概念分层工作,并在此基础上进行了关联规则的挖掘,包括单维关联规则的挖掘和多维关联规则的挖掘。结果表明建立在共有成分基础上的科的单维规则在Takhtaja系统中的解释度不如在Cronquist系统中的解释度。同时还挖掘出很多有趣的可以利用的关联规则比如降血糖(糖尿病)==>抗菌、如毛蓖科一>抗高血压药、抗高血压药:=>毛蓖科、茄科==>M胆碱受体拮抗剂、M胆碱受体拮抗剂==>茄科、安胎一>唇形科。 对中药化学数据库进行了包括主成分分析、因子分析、聚类分析以及判别分析等在内的数据挖掘工作。通过团子提取,获得了5个因子,得到一组分布更加密集,更有代表性的化合物集合,对因子进行了命名。考察了变量的选择和标准化对于聚类结果的影响。对一个5个活性的化合物集进行了判别分析,取得了较好的结果。对一系列的咪哇苯并二氮杂卓类化合物与五种重组GABAA受体亚型axβ3γ2(x=1,2,3,5,6)的亲和力进行了分子模型化的研究,得到了比较理想的模型,并对预报集中化合物的活性进行了预报,结果令人满意。比较分子场的研究表明:对于咪哇苯并二氮杂卓类化合物,CS位正电荷的小体积的取代基、C3位大体积的取代基有利于活性的提高。在FLARM方法中,同时建立了五个模型并得到了五个虚拟受体,并将虚拟受体和统一药效团/受体模型进行了比较发现模型具有一致性,同时说明CS位的取代对于a5的选择性起到很重要的作用。验证了CoMFA中C3位大体积的取代基有利于活性提高的结论,并对CoMFA和FLARM方法进行了比较。 |
英文摘要 | Data mining technology was introduced to the fields of traditional Chinese medicines and chemistry to mine in traditional Chinese medicines chemical database (TCMCD) after it was updated. Molecular modeling for the binding affinities of a series of imidazobenzodiazepines at five recombinant receptor subtypes was carried out successfully. The main contents and results were listed as follows: New version development of TCMCD was complished, which included standard expression set, database frame, structural information, TCM name system and validity examination, and several retrival examples were introduced in detailed, then function and usage of TCMCD were discussed. Statistics of "drug-like" characteristics of TCMCD showed molecular weight and LogP were in accordance with Lipinski rule. After data warehouse and data market were built and source data was cleaned, concept hierarchy was generated, then association rules included single-dimension and multidimension were mined. Results showed that Takhtaja system explains association rules of family better than Cronquist system. At the same time, many interesting and valuable rules such as hypoglycemic==> antibacterial, Ranunculaceae==> antihypertensive, antihypertensive==> Ranunculaceae, Solanaceae==> anticholinergic, anticholinergic==> Solanaceae, quiet fetus ==> Labiatae were obtained. Data mining including principle component analysis, factor analysis, clustering analysis and discriminant analysis in TCMCD was carried out. After five factors were extracted by factor analysis, a representative compound set was gained by removing outliers using five factors, and then these five factors were named. Influence of variable selection and standardization on clustering results was studied. Good result was gained after a discriminant analysis of a five-activity compound set was accomplished. Molecular modeling for the binding affinities of a series of imidazobenzodiazepines at five recombinant receptor subtypes a^n (x=l, 2, 3, 5, 6) was carried out and predictable models were achieved, which were tested by a test set. Results of Comparative Molecular Field (CoMFA) showed imidazobenzodiazepines substituted by more positive charge and smaller size group at position C8 and larger size group at position C3 would enhance ligand affinity at GABAA receptor. Five Pseudoreceptors built by FLARM were compared with united pharmacophore/receptor model, the result of which manifested that they were in correspondence with each other. At the same time, the pseudoreceptors illustrated that it was essential to the a5 selection of ligand substituted at position C8. The conclusion that arger size group at position C3 would enhance ligand affinity at GABAA receptor in CoMFA was also verified. Finally, CoMFA and FLARM were compared. |
语种 | 中文 |
公开日期 | 2013-09-16 |
页码 | 150 |
源URL | [http://ir.ipe.ac.cn/handle/122111/1405] ![]() |
专题 | 过程工程研究所_研究所(批量导入) |
推荐引用方式 GB/T 7714 | 陆爱军. 中药化学数据库的新版本、数据挖掘及GABA_A受体的分子模型化[D]. 中国科学院过程工程研究所. 中国科学院过程工程研究所. 2004. |
入库方式: OAI收割
来源:过程工程研究所
浏览0
下载0
收藏0
其他版本
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。