A hybrid approach to learn with imbalanced classes using evolutionary algorithms

There is an increasing interest in the application of Evolutionary Algorithms (EAs) to induce classification rules. This hybrid approach can benefit areas where classical methods for rule induction have not been very successful. One example is the induction of classification rules in imbalanced domains. Imbalanced data occur when one or more classes heavily outnumber other classes. Frequently, classical machine learning (ML) classifiers are not able to learn in the presence of imbalanced data sets, inducing classification models that always predict the most numerous classes. In this work, we propose a novel hybrid approach to deal with this problem. We create several balanced data sets with all minority class cases and a random sample of majority class cases. These balanced data sets are fed to classical ML systems that produce rule sets. The rule sets are combined creating a pool of rules and an EA is used to build a classifier from this pool of rules. This hybrid approach has some advantages over undersampling, since it reduces the amount of discarded information, and some advantages over oversampling, since it avoids overfitting. The proposed approach was experimentally analysed and the experimental results show an improvement in the classification performance measured as the area under the receiver operating characteristics (ROC) curve.

A pattern recognition approach to complex networks

Complex networks exist in many areas of science such as biology, neuroscience, engineering, and sociology. The growing development of this area has led to the introduction of several topological and dynamical measurements, which describe and quantify the structure of networks. Such characterization is essential not only for the modeling of real systems but also for the study of dynamic processes that may take place in them. However, it is not easy to use several measurements for the analysis of complex networks, due to the correlation between them and the difficulty of their visualization. To overcome these limitations, we propose an effective and comprehensive approach for the analysis of complex networks, which allows the visualization of several measurements in a few projections that contain the largest data variance and the classification of networks into three levels of detail, vertices, communities, and the global topology. We also demonstrate the efficiency and the universality of the proposed methods in a series of real-world networks in the three levels.

Time-resolved study electronic and thermal contributions to the nonlinear refractive index of Nd3+: SBN laser crystals

The propagation of an optical beam through dielectric media induces changes in the refractive index, An, which causes self-focusing or self-defocusing. In the particular case of ion-doped solids, there are thermal and non-thermal lens effects, where the latter is due to the polarizability difference, Delta alpha, between the excited and ground states, the so-called population lens (PL) effect. PL is a pure electronic contribution to the nonlinearity, while the thermal lens (TL) effect is caused by the conversion of part of the absorbed energy into heat. In time-resolved measurements such as Z-scan and TL transient experiments, it is not easy to separate these two contributions to nonlinear refractive index because they usually have similar response times. In this work, we performed time-resolved measurements using both Z-scan and mode mismatched TL in order to discriminate thermal and electronic contributions to the laser-induced refractive index change of the Nd3+-doped Strontium Barium Niobate (SrxBa1-xNb2O6) laser crystal. Combining numerical simulations with experimental results we could successfully distinguish between the two contributions to An. (C) 2007 Elsevier B.V. All rights reserved.

Electronic Structure and Spectro-Structural Correlations of Fe(III)Zn(II) Biomimetics for Purple Acid Phosphatases: Relevance to DNA Cleavage and Cytotoxic Activity

Purple acid phosphatases (PAPs) are a group of metallohydrolases that contain a dinuclear Fe(II)M(II) center (M(II) = Fe, Mn, Zn) in the active site and are able to catalyze the hydrolysis of a variety of phosphoric acid esters. The dinuclear complex [(H(2)O)Fe(III)(mu-OH)Zn(II)(L-H)](CIO(4))(2) (2) with the ligand 2-[N-bis(2-pyridylmethyl)aminomethyl]-4-methyl-6-[N-(2-pyridylmethyl)(2-hydroxybenzyl) aminomethyl]phenol (H(2)L-H) has recently been prepared and is found to closely mimic the coordination environment of the Fe(III)Zn(II) active site found in red kidney bean PAP (Neves et al. J. Am. Chem. Soc. 2007, 129, 7486). The biomimetic shows significant catalytic activity in hydrolytic reactions. By using a variety of structural, spectroscopic, and computational techniques the electronic structure of the Fe(III) center of this biomimetic complex was determined. In the solid state the electronic ground state reflects the rhombically distorted Fe(III)N(2)O(4) octahedron with a dominant tetragonal compression align ad along the mu-OH-Fe-O(phenolate) direction. To probe the role of the Fe-O(phenolate) bond, the phenolate moiety was modified to contain electron-donating or -withdrawing groups (-CH(3), -H, -Br, -NO(2)) in the 5-position. Tie effects of the substituents on the electronic properties of the biomimetic complexes were studied with a range of experimental and computational techniques. This study establishes benchmarks against accurate crystallographic struck ral information using spectroscopic techniques that are not restricted to single crystals. Kinetic studies on the hydrolysis reaction revealed that the phosphodiesterase activity increases in the order -NO(2)<- Br <- H <- CH(3) when 2,4-bis(dinitrophenyl)phosphate (2,4-bdnpp) was used as substrate, and a linear free energy relationship is found when log(k(cat)/k(0)) is plotted against the Hammett parameter a. However, nuclease activity measurements in the cleavage of double stranded DNA showed that the complexes containing the electron-withdrawing -NO(2) and electron-donating CH3 groups are the most active while the cytotoxic activity of the biomimetics on leukemia and lung tumoral cells is highest for complexes with electron-donating groups.

A New Approach to the Prediction of Partition Coefficients in Water/Organic Interfaces

In the present work, a new approach for the determination of the partition coefficient in different interfaces based on the density function theory is proposed. Our results for log P(ow) considering a n-octanol/water interface for a large super cell for acetone -0.30 (-0.24) and methane 0.95 (0.78) are comparable with the experimental data given in parenthesis. We believe that these differences are mainly related to the absence of van der Walls interactions and the limited number of molecules considered in the super cell. The numerical deviations are smaller than that observed for interpolation based tools. As the proposed model is parameter free, it is not limited to the n-octanol/water interface.

Saccharomyces cerevisiae coq10 null mutants are responsive to antimycin A

Deletion of COQ10 in Saccharomyces cerevisiae elicits a respiratory defect characterized by the absence of cytochrome c reduction, which is correctable by the addition of exogenous diffusible coenzyme Q(2). Unlike other coq mutants with hampered coenzyme Q(6) (Q(6)) synthesis, coq10 mutants have near wild-type concentrations of Q(6). In the present study, we used Q-cycle inhibitors of the coenzyme QH(2)-cytochrome c reductase complex to assess the electron transfer properties of coq10 cells. Our results show that coq10 mutants respond to antimycin A, indicating an active Q-cycle in these mutants, even though they are unable to transport electrons through cytochrome c and are not responsive to myxothiazol. EPR spectroscopic analysis also suggests that wild-type and coq10 mitochondria accumulate similar amounts of Q(6) semiquinone, despite a lower steady-state level of coenzyme QH(2)-cytochrome c reductase complex in the coq10 cells. Confirming the reduced respiratory chain state in coq10 cells, we found that the expression of the Aspergillus fumigatus alternative oxidase in these cells leads to a decrease in antimycin-dependent H(2)O(2) release and improves their respiratory growth.

Nanoparticle Platform to Modulate Reaction Mechanism of Phenothiazine Photosensitizers

Herein, we report on the synthesis of photosensitizing nanoparticles in which the generation of different oxidizing species, i.e., singlet oxygen ((1)O(2)) or radicals, was modulated. Sol gel and surface chemistry were used to obtain nanoparticles with specific ratios of dimer to monomer species of phenothiazine photosensitizers (PSs). Due to competition between the reactions involving electron transfer within dimer species and energy transfer from monomer triplets to oxygen, the efficiency of (1)O(2) generation could be controlled. Nanoparticles with an excess of dimer have an (1)O(2) generation efficiency (S(Delta)) of 0.01 while those without dimer have a S, value of 0.4. Furthermore, we demonstrate that the PS properties of the nanoparticles are not subjected to interference from the external medium as is commonly the case for free PSs, i.e., PS ground and triplet states are not reduced by NADH and ascorbate, respectively, and singlet excited states are less suppressed by bromide. The modulated (1)O(2) generation and the PS protection from external interferences make this nanoparticle platform a promising tool to aid in performing mechanistic studies in biological systems. Also, it offers potential application in technological areas in which photo-induced processes take place.

Introducing Undergraduate Students to Science

Understanding the scientific method fosters the development of critical thinking and logical analysis of information. Additionally, proposing and testing a hypothesis is applicable not only to science, but also to ordinary facts of daily life. Knowing the way science is done and how its results are published is useful for all citizens and mandatory for science students. A 60-h course was created to offer undergraduate students a framework in which to learn the procedures of scientific production and publication. The course`s main focus was biochemistry, and it was comprised of two modules. Module I dealt with scientific articles, and Module II with research project writing. Module I covered the topics: 1) the difference between scientific knowledge and common sense, 2) different conceptions of science, 3) scientific methodology, 4) scientific publishing categories, 5) logical principles, 6) deductive and inductive approaches, and 7) critical reading of scientific articles. Module II dealt with 1) selection of an experimental problem for investigation, 2) bibliographic revision, 3) materials and methods, 4) project writing and presentation, 5) funding agencies, and 6) critical analysis of experimental results. The course adopted a collaborative learning strategy, and each topic was studied through activities performed by the students. Qualitative and quantitative course evaluations with Likert questionnaires were carried out at each stage, and the results showed the students` high approval of the course. The staff responsible for course planning and development also evaluated it positively. The Biochemistry Department of the Chemistry Institute of the University of Sao Paulo has offered the course four times.