RESUMO
BACKGROUND: Recursive partitioning is a non-parametric modeling technique, widely used in regression and classification problems. Model-based recursive partitioning is used to identify groups of observations with similar values of parameters of the model of interest. The mob() function in the party package in R implements model-based recursive partitioning method. This method produces predictions based on single tree models. Predictions obtained through single tree models are very sensitive to small changes to the learning sample. We extend the model-based recursive partition method to produce predictions based on multiple tree models constructed on random samples achieved either through bootstrapping (random sampling with replacement) or subsampling (random sampling without replacement) on learning data. RESULTS: Here we present an R package called "mobForest" that implements bagging and random forests methodology for model-based recursive partitioning. The mobForest package constructs large number of model-based trees and the predictions are aggregated across these trees resulting in more stable predictions. The package also includes functions for computing predictive accuracy estimates and plots, residuals plot, and variable importance plot. CONCLUSION: The mobForest package implements a random forest type approach for model-based recursive partitioning. The R package along with it source code is available at http://CRAN.R-project.org/package=mobForest.
Assuntos
Modelos Estatísticos , Software , Algoritmos , Estatísticas não ParamétricasRESUMO
Population-based variability in protein expression patterns, especially in humans, is often observed but poorly understood. Moreover, very little is known about how interindividual genetic variation contributes to protein expression patterns. To begin to address this, we describe elements of technical and biological variations contributing to expression of 544 proteins in a population of 24 individual human lymphoblastoid cell lines that have been extensively genotyped as part of the International HapMap Project. We determined that expression levels of 10% of the proteins were tightly correlated to cell doubling rates. Using the publicly available genotypes for these lymphoblastoid cell lines, we applied a genetic association approach to identify quantitative trait loci associated with protein expression variation. Results identified 24 protein forms corresponding to 15 proteins for which genetic elements were responsible for >50% of the expression variation. The genetic variation associated with protein expression levels were located in cis with the gene coding for the transcript of the protein for 19 of these protein forms. Four of the genetic elements identified were coding non-synonymous single nucleotide polymorphisms that resulted in migration pattern changes in the two-dimensional gel. This is the first description of large scale proteomics analysis demonstrating the direct relationship between genome and proteome variations in human cells.
Assuntos
Variação Genética , Linfócitos/fisiologia , Proteoma/análise , Proteoma/genética , Locos de Características Quantitativas , Animais , Linhagem Celular , Eletroforese em Gel Bidimensional , Genótipo , Humanos , Linfócitos/citologiaRESUMO
Homeostasis in the immune system is maintained by specialized regulatory CD4(+) T cells (T(reg)) expressing transcription factor Foxp3. According to the current paradigm, high-affinity interactions between TCRs and class II MHC-peptide complexes in thymus "instruct" developing thymocytes to up-regulate Foxp3 and become T(reg) cells. However, the loss or down-regulation of Foxp3 does not disrupt the development of T(reg) cells but abrogates their suppressor function. In this study, we show that Foxp3-deficient T(reg) cells in scurfy mice harboring a null mutation of the Foxp3 gene retained cellular features of T(reg) cells including in vitro anergy, impaired production of inflammatory cytokines, and dependence on exogenous IL-2 for proliferation and homeostatic expansion. Foxp3-deficient T(reg) cells expressed a low level of activation markers, did not expand relative to other CD4(+) T cells, and produced IL-4 and immunomodulatory cytokines IL-10 and TGF-beta when stimulated. Global gene expression profiling revealed significant similarities between T(reg) cells expressing and lacking Foxp3. These results argue that Foxp3 deficiency alone does not convert T(reg) cells into conventional effector CD4(+) T cells but rather these cells constitute a distinct cell subset with unique features.
Assuntos
Linfócitos T CD4-Positivos/citologia , Fatores de Transcrição Forkhead/genética , Subpopulações de Linfócitos T/classificação , Linfócitos T Reguladores/citologia , Animais , Linfócitos T CD4-Positivos/imunologia , Proliferação de Células , Citocinas/biossíntese , Fatores de Transcrição Forkhead/deficiência , Perfilação da Expressão Gênica , Homeostase/imunologia , Ativação Linfocitária , Camundongos , Camundongos Mutantes , Linfócitos T Reguladores/imunologiaRESUMO
BACKGROUND: During the last decade, the use of microarrays to assess the transcriptome of many biological systems has generated an enormous amount of data. A common technique used to organize and analyze microarray data is to perform cluster analysis. While many clustering algorithms have been developed, they all suffer a significant decrease in computational performance as the size of the dataset being analyzed becomes very large. For example, clustering 10000 genes from an experiment containing 200 microarrays can be quite time consuming and challenging on a desktop PC. One solution to the scalability problem of clustering algorithms is to distribute or parallelize the algorithm across multiple computers. RESULTS: The software described in this paper is a high performance multithreaded application that implements a parallelized version of the K-means Clustering algorithm. Most parallel processing applications are not accessible to the general public and require specialized software libraries (e.g. MPI) and specialized hardware configurations. The parallel nature of the application comes from the use of a web service to perform the distance calculations and cluster assignments. Here we show our parallel implementation provides significant performance gains over a wide range of datasets using as little as seven nodes. The software was written in C# and was designed in a modular fashion to provide both deployment flexibility as well as flexibility in the user interface. CONCLUSION: ParaKMeans was designed to provide the general scientific community with an easy and manageable client-server application that can be installed on a wide variety of Windows operating systems.
Assuntos
Algoritmos , Inteligência Artificial , Análise por Conglomerados , Perfilação da Expressão Gênica/métodos , Família Multigênica/fisiologia , Análise de Sequência com Séries de Oligonucleotídeos/métodos , Reconhecimento Automatizado de Padrão/métodos , SoftwareRESUMO
Dendritic cells uniquely orchestrate the delicate balance between T cell immunity and regulation and an imbalance favoring immunogenic rather than tolerogenic DC is believed to contribute to the development of autoimmune diseases such as type 1 diabetes (T1D). In this study, we determined the frequencies of three blood DC subsets (pDC, mDC1 and mDC2) in 72 T1D patients and 75 normal controls using the Miltenyi blood DC enumeration kit. The frequency of blood pDC was found to be negatively correlated with subject age in both normal controls and T1D patients (p=0.0007), while the frequency of mDC1 and mDC2 do not change significantly with subject age. More importantly, the mean frequency of pDC in blood was, after adjusting for age, significantly lower in T1D (mean=0.127%) than controls (mean=0.188%) (p<6.0 x 10(-5)), whereas no difference was observed for mDC1 and mDC2 between T1D and controls. Furthermore, T1D patients have a lower proportion of pDC and higher proportion of mDC1 among the total blood DC population than normal controls. These results indicate that the frequency of blood pDC and the pDC/mDC1 ratio are negatively associated with T1D.
Assuntos
Células Dendríticas/imunologia , Diabetes Mellitus Tipo 1/imunologia , Adolescente , Adulto , Estudos de Casos e Controles , Contagem de Células , Criança , Pré-Escolar , Estudos de Coortes , Células Dendríticas/patologia , Diabetes Mellitus Tipo 1/sangue , Feminino , Citometria de Fluxo , Humanos , Lactente , Masculino , Pessoa de Meia-Idade , Adulto JovemRESUMO
MOTIVATION: In cluster analysis, the validity of specific solutions, algorithms, and procedures present significant challenges because there is no null hypothesis to test and no 'right answer'. It has been noted that a replicable classification is not necessarily a useful one, but a useful one that characterizes some aspect of the population must be replicable. By replicable we mean reproducible across multiple samplings from the same population. Methodologists have suggested that the validity of clustering methods should be based on classifications that yield reproducible findings beyond chance levels. We used this approach to determine the performance of commonly used clustering algorithms and the degree of replicability achieved using several microarray datasets. METHODS: We considered four commonly used iterative partitioning algorithms (Self Organizing Maps (SOM), K-means, Clutsering LARge Applications (CLARA), and Fuzzy C-means) and evaluated their performances on 37 microarray datasets, with sample sizes ranging from 12 to 172. We assessed reproducibility of the clustering algorithm by measuring the strength of relationship between clustering outputs of subsamples of 37 datasets. Cluster stability was quantified using Cramer's v2 from a kXk table. Cramer's v2 is equivalent to the squared canonical correlation coefficient between two sets of nominal variables. Potential scores range from 0 to 1, with 1 denoting perfect reproducibility. RESULTS: All four clustering routines show increased stability with larger sample sizes. K-means and SOM showed a gradual increase in stability with increasing sample size. CLARA and Fuzzy C-means, however, yielded low stability scores until sample sizes approached 30 and then gradually increased thereafter. Average stability never exceeded 0.55 for the four clustering routines, even at a sample size of 50. These findings suggest several plausible scenarios: (1) microarray datasets lack natural clustering structure thereby producing low stability scores on all four methods; (2) the algorithms studied do not produce reliable results and/or (3) sample sizes typically used in microarray research may be too small to support derivation of reliable clustering results. Further research should be directed towards evaluating stability performances of more clustering algorithms on more datasets specially having larger sample sizes with larger numbers of clusters considered.
Assuntos
Análise Serial de Proteínas/métodos , Análise Serial de Proteínas/normas , Análise por ConglomeradosRESUMO
Traumatic brain injury, depression and posttraumatic stress disorder (PTSD) are neurocognitive syndromes often associated with impairment of physical and mental health, as well as functional status. These syndromes are also frequent in military service members (SMs) after combat, although their presentation is often delayed until months after their return. The objective of this prospective cohort study was the identification of independent predictors of neurocognitive syndromes upon return from deployment could facilitate early intervention to prevent disability. We completed a comprehensive baseline assessment, followed by serial evaluations at three, six, and 12 months, to assess for new-onset PTSD, depression, or postconcussive syndrome (PCS) in order to identify baseline factors most strongly associated with subsequent neurocognitive syndromes. On serial follow-up, seven participants developed at least one neurocognitive syndrome: five with PTSD, one with depression and PTSD, and one with PCS. On univariate analysis, 60 items were associated with syndrome development at p < 0.15. Decision trees and ensemble tree multivariate models yielded four common independent predictors of PTSD: right superior longitudinal fasciculus tract volume on MRI; resting state connectivity between the right amygdala and left superior temporal gyrus (BA41/42) on functional MRI; and single nucleotide polymorphisms in the genes coding for myelin basic protein as well as brain-derived neurotrophic factor. Our findings require follow-up studies with greater sample size and suggest that neuroimaging and molecular biomarkers may help distinguish those at high risk for post-deployment neurocognitive syndromes.
RESUMO
The microbiome is being characterized by large-scale sequencing efforts, yet it is not known whether it regulates host metabolism in a general versus tissue-specific manner or which bacterial metabolites are important. Here, we demonstrate that microbiota have a strong effect on energy homeostasis in the colon compared to other tissues. This tissue specificity is due to colonocytes utilizing bacterially produced butyrate as their primary energy source. Colonocytes from germfree mice are in an energy-deprived state and exhibit decreased expression of enzymes that catalyze key steps in intermediary metabolism including the TCA cycle. Consequently, there is a marked decrease in NADH/NAD(+), oxidative phosphorylation, and ATP levels, which results in AMPK activation, p27(kip1) phosphorylation, and autophagy. When butyrate is added to germfree colonocytes, it rescues their deficit in mitochondrial respiration and prevents them from undergoing autophagy. The mechanism is due to butyrate acting as an energy source rather than as an HDAC inhibitor.
Assuntos
Autofagia , Butiratos/farmacologia , Colo/metabolismo , Metabolismo Energético , Metagenoma , Proteínas Quinases Ativadas por AMP/metabolismo , Animais , Biomarcadores/metabolismo , Western Blotting , Células Cultivadas , Colo/citologia , Inibidor de Quinase Dependente de Ciclina p27/metabolismo , Perfilação da Expressão Gênica , Vida Livre de Germes , Glucose/metabolismo , Espectroscopia de Ressonância Magnética , Masculino , Metabolômica , Camundongos , Camundongos Endogâmicos C57BL , Mitocôndrias/metabolismo , NAD/metabolismo , Análise de Sequência com Séries de Oligonucleotídeos , Fosforilação Oxidativa , Fosforilação , Transdução de SinaisRESUMO
BACKGROUND: CD8+ NKT-like cells are naturally occurring but rare T cells that express both T cell and natural killer cell markers. These cells may play key roles in establishing tolerance to self-antigens; however, their mechanism of action and molecular profiles are poorly characterized due to their low frequencies. We developed an efficient in vitro protocol to produce CD8+ T cells that express natural killer cell markers (CD8+ NKT-like cells) and extensively characterized their functional and molecular phenotypes using a variety of techniques. RESULTS: Large numbers of CD8+ NKT-like cells were obtained through culture of naïve CD8+ T cells using anti-CD3/anti-CD28-coated beads and high dose IL-2. These cells possess potent activity in suppressing the proliferation of naïve responder T cells. Gene expression profiling suggests that the cultured CD8+ NKT-like cells and the naïve CD8+ T cells differ by more than 2-fold for about 3,000 genes, among which 314 are upregulated by more than 5-fold and 113 are upregulated by more than 10-fold in the CD8+ NKT-like cells. A large proportion of the highly upregulated genes are soluble factors or surface markers that have previously been implicated in immune suppression or are likely to possess immunosuppressive properties. Many of these genes are regulated by two key cytokines, IL-10 and IFN-gamma. The immunosuppressive activities of cells cultured from IL-10-/- and IFN-gamma-/- mice are reduced by about 70% and about 50%, respectively, compared to wild-type mice. CONCLUSION: Immunosuppressive CD8+ NKT-like cells can be efficiently produced and their immunosuppressive activity is related to many surface and soluble molecules regulated by IL-10 and IFN-gamma.