基于启发式方法和支持向量机方法预测有机物的热导率

构建147个有机物分子结构与其热导率值之间的定量结构-性质关系(QSPR)模型, 探讨影响有机物热导率的结构因素. 以147个化合物作为样本集, 随机选择118个作为训练集, 29个作为测试集. 应用CODESSA软件计算了组成、拓扑、几何、静电和量子化学等描述符, 通过启发式方法(HM)筛选得到5个结构参数并建立线性回归模型; 用所选5个结构参数作为支持向量机(SVM)的输入, 建立非线性的支持向量机回归模型. 预测结果表明: 支持向量机回归模型的性能(复相关系数R²=0.9240)虽略低于启发式回归模型的性能(R²=0.9267), 但是支持向量机方法预测性能(R²=0.9682)高于启发式方法的预测性能(R²=0.9574), 对于QSPR模型来说, 预测性能更重要. 因此, 总体来说支持向量机方法优于启发式方法. 支持向量机方法和启发式方法的提出为工程上提供了一种根据分子结构预测有机物热导率的新方法.     

基于启发式方法和支持向量机方法预测药物与人血浆蛋白结合率

应用启发式方法和支持向量机方法建立了70种药物与血浆蛋白结合率的定量构效关系模型, 研究了分子结构对药物与血浆蛋白结合率的影响. 两种方法均得到了较好的结果, 交互检验的相关系数平方分别为0.80和0.82; 通过对模型的稳定性和预测能力比较, 支持向量机建立的QSAR模型能够更好地预测药物与血浆蛋白结合率.     

最小二乘支持向量机方法用于提高低水平量子化学方法计算吸收能的精度

运用B3LYP/STO-3G和ZINDO两种低水平的量子化学方法计算了160个有机分子的UV-Vis吸收光谱, 然后提取合适的物理参数, 并以实验值为基础, 引入最小二乘支持向量机方法以提高吸收能的计算值精度. 结果表明, 最小二乘支持向量机方法可有效提高量子化学计算精度, 体系的吸收能误差均方根分别从0.95和0.46 eV降低到0.16和0.15 eV. 最小二乘支持向量机校正方法的引入可在较少的机时和计算资源下得到比单一的量子化学计算方法更为稳定和精确的计算结果, 且可在现有计算条件下预测现有计算能力达不到的精度. 因此, 将最小二乘支持向量机方法用于量子化学数据分析, 为化学研究准确、 快捷地预测分子性质提供了一种新的研究手段.     

氨基酸0D-3D信息得分矢量用于人免疫缺陷病毒蛋白酶裂解位点预测及特异性分析

收集天然氨基酸的1369种0D-3D结构信息参数,经主成分分析得一组新氨基酸描述子——氨基酸0D-3D信息得分矢量,将其用于人免疫缺陷病毒蛋白酶（HIV PR）裂解位点预测,以线性判别分析与支持向量机建模预测HIV PR裂解位点．线性判别分析与支持向量机模型对646个训练集样本的自检验识别、留一法交互验证及对100个测试集样本外部验证的马休斯相关系数分别为0．879和0．911,0．849和0．901,0．822和0．846．经受试者操作特征曲线分析表明,支持向量机对HIVPR裂解位点的预测结果优于线性判别分析．研究显示,氨基酸0D-3D信息得分矢量可进一步用于HIVPR裂解位点预测．     

支持向量机分类和回归用于肽的QSAR研究

使用支持向量机技术对两类肽化合物体系进行了分类和回归研究,并将其系统地与K最邻近法、多元线性回归、偏最小二乘、人工神经网络进行了比较。结果表明,对于小样本、非线性问题,支持向量机具有较强的稳定性能及泛化能力,在大多数情况下能够得到优于传统方法的建模效果。对于分类问题,支持向量机对训练集和测试集都达到了100%的分类正确率;对于回归问题,支持向量机虽对训练集样本拟合效果略低于人工神经网络,但对外部测试集却表现出较强的预测能力。     

七种无机元素与中药药性的相关性研究

本文研究支持向量机用于中药药性识别的可行性.选择7种无机元素在中药中的含量为指标,运用支持向量机对193种不同药性的中药进行训练,建立平性与非平性中药的预测模型.结果训练集识别正确率95.03%,建立的模型对预测集中平性药的识别正确率为82.14%,对非平性药的识别正确率为70%,总正确率为73.58%.     

基于机器学习方法的激素敏感脂肪酶抑制剂活性预测(英文)

脂肪组织中,激素敏感脂肪酶(HSL)被认为是调节脂肪酸代谢的关键限速酶.HSL在糖尿病的发病过程中起重要作用,抑制HSL活性有助于糖尿病的治疗,因此探索新颖的HSL抑制剂成为当前研究的热门.在激素敏感脂肪酶的作用机制和三维结构缺乏的情况下,需要发展预测HSL抑制剂的方法.本文采用几种机器学习方法(支持向量机(SVM)、k-最近相邻法(k-NN)和C4.5决策树(C4.5DT))对已知的HSL抑制剂与非抑制剂建立分类预测模型.252个结构多样性化合物(123个HSL抑制剂与129个HSL非抑制剂)被用于测试分类预测系统,并用递归变量消除法选择与HSL抑制剂相关的性质描述符以提高预测精度.本研究对独立验证集的总预测精度为75.0%-80.0%,HSL抑制剂的预测精度为85.7%-90.5%,非HSL抑制剂的预测精度为63.2%-68.4%.支持向量机方法给出最好的总预测精度(80.0%).本研究表明支持向量机等机器学习方法可以有效预测未知数据集中潜在的HSL抑制剂,并有助于发现与其相关的分子描述符.     

主成分分析-支持向量机用于肝病分类模型

将主成分分析(PCA)用于肝功能检测数据特征提取,然后用支持向量机(SVM)对乙肝、丙肝、肝硬化、正常人样本建立分类模型。采用高斯径向基函数(RBF)为核函数,调节核函数参数C及σ以建立最佳支持向量机模型。该模型对训练集的识别率为99.3%,对预测集的预测率为96.4%。结果表明:PCA-SVM法建立的肝病分类模型能较好的区分乙肝、丙肝、肝硬化及正常人,且分类效果优于传统支持向量机及人工神经网络(ANN)分类模型。     

杭州老虎洞窑古陶瓷成分的化学计量学研究

用支持向量机算法研究了与杭州老虎洞古陶瓷有关的两个断源、断代问题。作为化学计量学的～种新型分类算法,支持向量机在小样本问题上表现出良好的泛化能力,与特征选择方法结合,可以有效处理样本少,特征多的问题。本研究综合利用支持向量机、特征选择算法和其它化学计量学算法研究了杭州凤凰山麓万松岭附近的古窑遗址和“传世哥窑”的断源、断代问题,证明老虎洞窑与郊坛下窑产品截然不同,万松岭附近地面收集瓷片样本是老虎洞窑宋代地层的瓷片滑落所致,而“传世哥窑”样品可能是老虎洞窑元代时的产品。实验表明：支持向量机算法与化学分析相结合可以成为研究古陶瓷断源和断代问题的一种新方法。     

基于启发式算法和支持向量机算法预测醛类化合物急性毒性

应用启发式算法(HM)和支持向量机算法(SVM)建立了88种醛类化合物对大鼠急性毒性的定量构效关系模型。应用ChemOffice2004软件进行化合物的结构输入,利用半经验方法进行分子结构优化,在CODDESA软件中计算出组成、拓扑、几何、电子和量子化学参数。并用启发式方法筛选出相对阳性电荷、氧原子数量、碳原子的最小亲核反应指数、双键相对数量、碳原子数量、碳氢键之间的最大交换能量、最大σ-σ键序和双键数量8个参数,在此基础上应该多元线性回归和支持向量机方法建立QSPR模型。两种方法均得到了较好的结果,HM和SVM的交互检验的相关系数分别为0.90和0.93;通过对模型的稳定性和预测能力比较,SVM建立的QSAR模型能够更好地预测醛类化合物对大鼠急性毒性LD50。     

二氧化碳和丙烯直接合成甲基丙烯酸的NiPMo/TiO2催化剂的研究

用溶胶-凝胶法以磷钼酸(MPA)的镍盐溶液水解钛酸四丁酯制备了NiPMo/TiO2催化剂.使用ICP、 XRD、 TG-DTA、 IR、 TPD-MS和微反应技术研究了催化剂的化学组成、热稳定性、化学吸附性质和催化反应性能.杂多钼酸盐与TiO2通过O2-在TiO2表面发生了键合.在623 K下,杂多阴离子仍保持原有的Keggin结构.CO2在Lewis酸位Ni(Ⅱ)和Lewis碱位Ni-O-Mo的桥氧协同作用下生成CO2卧式吸附态Ni(Ⅱ)←O-(CO)←(O--Ni).丙烯有多种吸附态在催化剂上吸附.在563 K、 1 MPa和空速1500 h-1的反应条件下,丙烯的摩尔转化率为3.2%,产物MAA选择性为95%.     

Synthesis of 1-benzyloxypyrazin-2(1H)-one derivatives

Different approaches for the synthesis of 1-benzyloxypyrazin-2(1H)-one derivatives from simple amino acids have been investigated. A library of 33 precursors for the preparation of N-hydroxy pyrazinones was obtained in moderate to good yields.     

Toward new camptothecins. Part 6: Synthesis of crucial ketones and their use in Friedländer reaction

In the context of the preparation of camptothecin and luotonin A analogs, the synthesis of some key keto-precursors and their use in Friedländer condensation are described. This paper also focuses on the stability of these keto intermediates and emphasizes the major differences between indolizinones and pyrroloquinazolinones series. Noteworthy is also the report of some original structures isolated as by-products of some experiments.     

CoFe2O4自形成磁性液体场致结构化对磁化的影响

The Langevin paramagnetic theory can’t describe the relation between magnetization of ferrofluids and applied magnetic field. The structuralization of ferrofluids, which is considered the main influence factor of the magnetization, is regarded. The part of magnetization works is deposited when the structure is forming. This action influences the magnetization of ferrofluids directly or indirectly. On the base of the “compressing” model, the Langevin function that usually describes the magnetization of ferrofluid is modified, and a well-fitted curve is obtained. An equation of the relation between the equivalent volume fraction after being “compressed” and the intensity of magnetic field is discovered, which approximately describes the process of magnetization. The relation between the approximate initial susceptibility and the volume fraction can be obtained from modified formula.     

Highly regioselective Buchwald–Hartwig amination at C-2 of 2,4-dichloropyridine enabling a novel approach to 2,4-bisanilinopyridine (BAPyd) libraries

The highly regioselective Buchwald–Hartwig amination at C-2 of the cheap and readily accessible reagent, 2,4-dichloropyridine with a range of anilines and heterocyclic amines is described. This new methodology is robust and provides a facile access to 4-chloro-N-phenylpyridin-2-amines on 0.25 mol scale. These intermediates undergo a further Buchwald–Hartwig amination at higher temperature to enable rapid exploration of the chemical space at C-4 and to provide a library of 2,4-bisaminopyridines.     

KMnO4-mediated oxidative CN bond cleavage of tertiary amines: Synthesis of amides and sulfonamides

KMnO₄-mediated oxidative CN bond cleavage of tertiary amines producing secondary amine was introduced, which was trapped by electrophiles (acyl chloride and sulfonyl chloride) to form amides and sulfonamides. The reaction could take place at mild condition, tolerating a wide range of function groups and affording products in moderate to excellent yields.     

Chemistry of heterocyclic compounds at the A. E. Favorsky Irkutsk Institute of Chemistry, Siberian Branch of the Russian Academy of Sciences, over 50 years (Review)

The review contains a concise historical account and information on the most significant researches undertaken by the staff at the A. E. Favorsky Irkutsk Institute of Chemistry, Siberian Branch of the Russian Academy of Sciences on the Chemistry of Heterocyclic Compounds. Dedicated to Academician of the Russian Academy of Sciences B. A. Trofimov on his 70th jubilee. Translated from Khimiya Geterotsiklicheskikh Soedinenii, No. 10, pp. 1443–1502, October, 2008.     

A general and convenient synthesis of 4-(tosylmethyl)semicarbazones and their use in amidoalkylation of hydrogen,heteroatom, and carbon nucleophiles

A general synthesis of previously unknown semicarbazone-based α-amidoalkylating reagents, 4-(tosylmethyl)semicarbazones, has been developed. The synthesis involved three-component condensation of semicarbazones of aliphatic or aromatic aldehydes with the same or other aldehydes and p-toluenesulfinic acid. The scope and limitations of this reaction were investigated. The compounds obtained were demonstrated to be an efficient α-(4-semicarbazono)alkylating agents. They were reacted with H- (sodium borohydride), O- (sodium methylate), S- (sodium phenylthiolate), N- (pyrrolidine, sodium succinimide), P- (trialkyl phosphites), and C-nucleophiles (sodium diethyl malonate) to give the corresponding products of the tosyl group substitution, 4-substituted semicarbazones, including analogues of nitrofurazone. Among the prepared compounds tested in vitro for antibacterial and antifungal activity, three nitrofuryl-containing semicarbazones exhibited high biological activities with minimum inhibitory concentration (MIC) values of 8–32 μg/mL.     

Synthesis of novel chiral bidentate hydroxyalkyl-N-heterocyclic carbene ligands and their application in palladium-catalyzed Mizoroki–Heck couplings and asymmetric addition of diethylzinc to benzaldehyde

A small library of new chiral bidentate hydroxyalkyl-imidazolium salts 1 is conveniently synthesized on multi-gram scale from inexpensive and commercially available chiral pool amino acids. The corresponding carbenes, generated by deprotonation of imidazolium salts 1, in combination with palladium(II) chloride were tested in the Mizoroki–Heck coupling reaction. The most significant results in terms of yields and reactivities were achieved with low catalyst loading. The catalytic activities of these imidazolium salts were also investigated in the asymmetric addition of diethylzinc to benzaldehyde. The use of MgO nanoparticles as an additive in conjunction with these ligands played a crucial role in increasing the efficiency of these reactions.     

Sulfur-directed metal-free and regiospecific methyl C(sp3)–H imidation of thioanisoles

Regiospecific and direct imidation of the methyl C(sp³)–H bond of thioanisoles is realized under mild and metal-free conditions with N-fluorobis(benzenesulfonyl)imide as an oxidant and nitrogen source. Proposed mechanism suggests that thionium ion intermediates and a Pummerer-type reaction are involved. The imidation has advantages such as high step-economy, excellent functionality tolerance, and regiospecificity, giving structurally diverse imidation products.