Large-scale regulatory network analysis from microarray data: modified Bayesian network learning and association rule mining

We present two algorithms for learning large-scale gene regulatory networks from microarray data: a modified information-theory-based Bayesian network algorithm and a modified association rule algorithm. Simulation-based evaluation using six datasets indicated that both algorithms outperformed their unmodified counterparts, especially when analyzing large numbers of genes. Both algorithms learned about 20% (50% if directionality and relation type were not considered) of the relations in the actual models. In our empirical evaluation based on two real datasets, domain experts evaluated subsets of learned relations with high confidence and identified 20–30% to be “interesting” or “maybe interesting” as potential experiment hypotheses.     

Class prediction and gene selection for DNA microarrays using regularized sliced inverse regression

The monitoring of the expression profiles of thousands of genes have proved to be particularly promising for biological classification. DNA microarray data have been recently used for the development of classification rules, particularly for cancer diagnosis. However, microarray data present major challenges due to the complex, multiclass nature and the overwhelming number of variables characterizing gene expression profiles. A regularized form of sliced inverse regression (REGSIR) approach is proposed. It allows the simultaneous development of classification rules and the selection of those genes that are most important in terms of classification accuracy. The method is illustrated on some publicly available microarray data sets. Furthermore, an extensive comparison with other classification methods is reported. The REGSIR performance is comparable with the best classification methods available, and when appropriate feature selection is made the performance can be considerably improved.     

Direct integration of microarrays for selecting informative genes and phenotype classification

The ability to provide thousands of gene expression values simultaneously makes microarray data very useful for phenotype classification. A major constraint in phenotype classification is that the number of genes greatly exceeds the number of samples. We overcame this constraint in two ways; we increased the number of samples by integrating independently generated microarrays that had been designed with the same biological objectives, and reduced the number of genes involved in the classification by selecting a small set of informative genes. We were able to maximally use the abundant microarray data that is being stockpiled by thousands of different research groups while improving classification accuracy. Our goal is to implement a feature (gene) selection method that can be applicable to integrated microarrays as well as to build a highly accurate classifier that permits straightforward biological interpretation. In this paper, we propose a two-stage approach. Firstly, we performed a direct integration of individual microarrays by transforming an expression value into a rank value within a sample and identified informative genes by calculating the number of swaps to reach a perfectly split sequence. Secondly, we built a classifier which is a parameter-free ensemble method using only the pre-selected informative genes. By using our classifier that was derived from large, integrated microarray sample datasets, we achieved high accuracy, sensitivity, and specificity in the classification of an independent test dataset.     

Local likelihood regression in generalized linear single-index models with applications to microarray data

Searching for an effective dimension reduction space is an important problem in regression, especially for high-dimensional data such as microarray data. A major characteristic of microarray data consists in the small number of observations n and a very large number of genes p. This “large p, small n” paradigm makes the discriminant analysis for classification difficult. In order to offset this dimensionality problem a solution consists in reducing the dimension. Supervised classification is understood as a regression problem with a small number of observations and a large number of covariates. A new approach for dimension reduction is proposed. This is based on a semi-parametric approach which uses local likelihood estimates for single-index generalized linear models. The asymptotic properties of this procedure are considered and its asymptotic performances are illustrated by simulations. Applications of this method when applied to binary and multiclass classification of the three real data sets Colon, Leukemia and SRBCT are presented.     

Diversity of Salmonella enterica serovar Derby isolated from pig, pork and humans in Germany

Salmonella enterica serovar Derby (S. Derby) is one of the most prevalent serovars in pigs in Europe and in the U.S. and ranks among the 10 most frequently isolated serovars in humans. Therefore, a set of 82 epidemiologically unrelated S. Derby strains isolated between 2006 and 2008 from pigs, pork and humans in Germany was selected and investigated in respect to the transmission of clonal groups of the serovar along the food chain. Various phenotypic and genotypic methods were applied and the pathogenicity and resistance gene repertoire was determined. Phenotypically 72% of the strains were susceptible to all 17 antimicrobials tested while the others were monoresistant to tetracycline or multi-resistant with different resistance profiles. Four major clonal groups were identified based on PFGE, sequence data of the virulence genes sopA, sopB and sopD, VNTR-locus STTR5 and MLST revealing also the new sequence type ST774. Thirty different PFGE profiles were detected resulting in four clusters representing the four groups. The pathogenicity gene repertoire of 32 representative S. Derby strains analyzed by microarray showed six types with differences in the Salmonella pathogenicity islands, pathogenicity genes on smaller islets or prophages and fimbriae coding genes. The pathogenicity gene repertoire of the predominant types PAT DE1 and DE2 were most similar to the ones of S. Paratyphi B (dT+, O5−) and to a minor degree to S. Infantis and S. Virchow PATs. Overall this study showed that in Germany currently one major S. Derby clone is frequently isolated from pigs and humans. Contaminated pork was identified as one vehicle and consequently is a risk for human health. To prevent this serovar from entering the food chain, control measurements should be applied at the farm level.     

Antibiotic susceptibility of members of the Lactobacillus acidophilus group using broth microdilution and molecular identification of their resistance determinants

The range of antibiotic susceptibility to 13 antibiotics in 101 strains of the Lactobacillus acidophilus group was examined using the lactic acid bacteria susceptibility test medium (LSM) and broth microdilution. Additionally, microarray analysis and PCR were applied to identify resistance genes responsible for the displayed resistant phenotypes in a selection of strains. In general, narrow as well as broad unimodal and bimodal MIC distributions were observed for the Lactobacillus acidophilus group and the tested antimicrobial agents. Atypically resistant strains could be determined by visual inspection of the obtained MIC ranges for ampicillin, chloramphenicol, clindamycin, erythromycin, quinupristin/dalfopristin, streptomycin and tetracycline. For most of these atypically resistant strains underlying resistance determinants were found. To our knowledge erm(A) was detected in lactobacilli for the first time within this study. Data derived from this study can be used as a basis for reviewing present microbiological breakpoints for categorization of susceptible and resistant strains within the Lactobacillus acidophilus group to assess the safety of microorganisms intended for use in food and feed applications.     

基于过采样技术和随机森林的不平衡微阵列数据分类方法研究

近年来,应用DNA微阵列技术对疾病,尤其是癌症进行诊断,已逐渐成为生物信息学领域的研究热点之一。对比其它的数据载体,微阵列数据通常具有一些独有的特点。针对微阵列数据样本分布不平衡这一特点,提出了一种基于概率分布的过采样技术,通过该技术可以为少数类建立一些合理的伪样本,从而使各类的样本数达到均衡,然后使用随机森林分类器对其进行分类。该方法的有效性和可行性已经在两个标准的微阵列数据集上得到了验证。实验结果显示,与传统的方法相比,该方法可以获得更好的分类性能。     

The effect of microarray image compression on expression-based classification

Current gene-expression microarrays carry enormous amounts of information. Compression is necessary for efficient distribution and storage. This paper examines JPEG2000 compression of cDNA microarray images and addresses the accuracy of classification and feature selection based on decompressed images. Among other options, we choose JPEG2000 because it is the latest international standard for image compression and offers lossy-to-lossless compression while achieving high lossless compression ratios on microarray images. The performance of JPEG2000 has been tested on three real data sets at different compression ratios, ranging from lossless to 45:1. The effects of JPEG2000 compression/decompression on differential expression detection and phenotype classification have been examined. There is less than a 4% change in differential detection at compression rates as high as 20:1, with detection accuracy suffering less than 2% for moderate to high intensity genes, and there is no significant effect on classification at rates as high as 35:1. The supplementary material is available at .     

Contrasting effects of fresh and fermented kimchi consumption on gut microbiota composition and gene expression related to metabolic syndrome in obese Korean women

Accumulating evidence suggests relationship of compositional changes of gut microbiota with onset of metabolic disorders and obesity. Kimchi, a traditional Korean side dish, is known for its beneficial impact on metabolic parameters and anti‐obesity effects. The current study was designed to evaluate the association between gut microbiota and human genome after kimchi intervention in an effort to understand the molecular mechanism(s) underlying the antiobesity impact of kimchi. Twenty‐four obese women were randomly assigned to either fresh or fermented kimchi group for eight weeks of kimchi intervention. Pyrosequencing of fecal microbiota and microarray analyses of blood samples revealed that fresh and fermented kimchi interventions exerted differential effects on the obesity‐related clinical parameters. Correlations of these effects with changes in blood gene expression and gut microbial population were more evident in the fermented kimchi group than the fresh kimchi group.