On estimating model complexity and prediction errors in multivariate calibration: generalized resampling by random sample weighting (RSW)

The present paper focuses on determining the number of PLS components by using resampling methods such as cross validation (CV), Monte Carlo cross validation (MCCV), bootstrapping (BS), etc. To resample the training data, random non‐negative weights are assigned to the original training samples and a sample‐weighted PLS model is developed without increasing the computational burden much. Random weighting is a generalization of the traditional resampling methods and is expected to have a lower risk of getting an insufficient training set. For prediction, only the training samples with random weights less than a threshold value are selected to ensure that the prediction samples have less influence on training. For complicated data, because the optimal number of PLS components is often not unique or readily distinguished and there might exist an optimal region of model complexity, the distribution of prediction errors can be more useful than a single value of root mean squared error of prediction (RMSEP). Therefore, the distribution of prediction errors are estimated by repeated random sample weighting and used to determine model complexity. RSW is compared with its traditional counterparts like CV, MCCV, BS and a recently proposed randomization test method to demonstrate its usefulness. Copyright © 2010 John Wiley & Sons, Ltd.     

Molecular Dynamics Simulation of HIV-1 Integrase Dimer Complexed with Viral DNA

The biological function of HIV‐1 integrase (IN) is to integrate viral DNA into the host cell chromosome, and the specific binding of IN with viral DNA is a precondition for IN to function correctly. Beforehand, the binding mode of IN dimer (IN₂) with the 27 bp segment of viral DNA before 3′ processing (3′‐P) was obtained via a molecular docking method. Based on the binding mode, the aim of this article was to explore the changes of motive mode and correlative motion for the IN₂ and DNA systems after their binding through dynamical cross‐correlation map (DCCM) and principal component analysis (PCA). Finally, solvent effect during the association was analyzed briefly. The results show that there is a significantly increased positive correlation in the interface region between IN₂ and viral DNA, and some obvious motive mode changes of the two systems (IN₂ and DNA) were also observed after their binding. It was found that water molecules played an important role in the recognition between IN₂ and viral DNA through analyzing the water‐mediated hydrogen bonds.     

Synthesis of the Phenylpyridal Scaffold as a Helical Peptide Mimetic

Phenylpyridal‐ and phenyldipyridal‐based scaffolds have been designed and synthesized as novel helical peptide mimetics. The synthesis required optimisation and selective alkylation in producing 2,6‐functionalized 3‐hydroxypyridine derivatives for a convergent scheme. The pyridine analogues were coupled by a series of Suzuki/Stille types cross‐coupling reactions. A series of biaryl and ter‐aryl substituted heterocycles were produced. The synthetic approach was concise and high yielding allowing large variability at the wanted side‐chain attachment points. A number of compounds were synthesised to show the versatility of the strategy.     

Transition‐Metal‐Free Formal Sonogashira Coupling and α‐Carbonyl Arylation Reactions

Transition‐metal‐free formal Sonogashira coupling and α‐carbonyl arylation reactions have been developed. These transformations are based on the nucleophilic aromatic substitution (S_NAr) of β‐carbonyl sulfones to electron‐deficient aryl fluorides, producing a key intermediate that, depending on the reaction conditions, gives the aromatic alkynes or α‐aryl carbonyl compounds. The development of these reactions is presented and, based on investigations under basic and acidic conditions, mechanisms have been proposed. To develop the formal Sonogashira coupling further, a milder, two‐step protocol is also disclosed that expands the reaction concept. The scope of these reactions is demonstrated for the synthesis of Sonogashira and α‐carbonyl arylated products from a range of electron‐deficient aryl fluorides with a variety of functional groups and aryl‐, heteroaryl‐, alkyl‐, and alkoxy‐substituted sulfone nucleophiles. These transition‐metal‐free reactions complement the metal‐catalyzed versions in terms of substitution patterns, simplicity, and reaction conditions.     

Detection of bio-constituents in complex biological tissue using Raman microscopy. Application to human nail clippings

Raman spectra of human nail clippings from various sources were collected and then deconvoluted to obtain the pure component spectra of the underlying constituents present. This blind-deconvolution was performed using a self-modeling curve resolution technique, namely band-target entropy minimization (BTEM). The aim was to simplify the complexity of the Raman spectra and hence to identify the underlying biological molecules in more detail. BTEM analysis could recover 13 pure component Raman spectral estimates from the collected 438 spectra measured from 113 nail samples. Six recovered pure component spectral estimates correspond to proteins or polypeptides that contain various amino acids such as phenylalanine, tyrosine, tryptophan, and cysteine. Two are associated with the secondary structures of proteins, and five are associated with two carotenoid species, lipid, ferulic acid, and calcium phosphate. Subsequently, the relative concentrations of these bio-constituents were calculated from the measured mixture spectra and the pure component BTEM estimates. These profiles indicated that the concentrations of some bio-constituents are correlated while others are not. A further analysis using target transformation factor analysis (TTFA) revealed the possible presence of curcumin in the human nails. Since the present approach and analysis is rather general, it might be extended to many other biological tissues in a rather straightforward and similar manner, thus revealing more detailed underlying biochemical information such as biomarkers that may be useful for diagnostic purposes.     

基于I-ELM的飞机复合材料结构损伤识别研究

针对飞机复合材料结构损伤难以有效识别问题,本文提出一种基于增量型极限学习机(Incremental Extreme Learning Machine,I-ELM)的飞机复合材料结构损伤识别新方法。首先对某型机用复合材料层合板进行冲击,而后对其进行疲劳拉伸试验,通过优化布局在复合材料层合板上的光纤光栅传感器募集应变信息,并对其进行预处理。采用互补总体平均经验模态分解(Complementary Ensemble Empirical Mode Decomposition,CEEMD)方法对募集的应变信息进行自适应分解,得到多个本征模态分量(Intrinsic Mode Function,IMF),计算各阶IMF分量的样本熵,通过核熵成分分析(Kernel Entropy Component Analysis,KECA)方法对样本熵进行特征融合,构建融合特征向量,采用融合特征向量建立基于I-ELM损伤识别模型,通过实验对损伤识别模型的有效性进行了验证,并与所构建的BP的损伤识别模型的识别结果进行了比较。结果表明,该方法能有效对飞机复合材料结构损伤进行识别,具有很好的应用前景。