Combining Supervised and Unsupervised Learning for Improved miRNA Target Prediction

MicroRNAs (miRNAs) are short non-coding RNAs which bind to mRNAs and regulate their expression. MiRNAs have been found to be associated with initiation and progression of many complex diseases. Investigating miRNAs and their targets can thus help develop new therapies by designing anti-miRNA oligonucleotides. While existing computational approaches can predict miRNA targets, these predictions have low accuracy. In this paper, we propose a two-step approach to refine the results of sequence-based prediction algorithms. The first step, which is based on our previous work, uses an ensemble learning approach that combines multiple existing methods. The second step utilizes support vector machine (SVM) classifiers in one- and two-class modes to infer miRNA-mRNA interactions based on both binding features, as well as network features extracted from gene regulatory network. Experimental results using two real data sets from TCGA indicate that the use of two-class SVM classification significantly improves the precision of miRNA-mRNA prediction.

Circulating bile acids in healthy adults respond differently to a dietary pattern characterized by whole grains, legumes and fruits and vegetables compared to a diet high in refined grains and added sugars: A randomized, controlled, crossover feeding study

OBJECTIVE

The effects of diets high in refined grains on biliary and colonic bile acids have been investigated extensively. However, the effects of diets high in whole versus refined grains on circulating bile acids, which can influence glucose homeostasis and inflammation through activation of farnesoid X receptor (FXR) and G protein-coupled bile acid receptor 1 (TGR5), have not been studied.

MATERIALS AND METHODS

We conducted a secondary analysis from a randomized controlled crossover feeding trial (NCT00622661) in 80 healthy adults (40 women/40 men, age 18-45 years) from the greater Seattle Area, half of which were normal weight (BMI 18.5-25.0 kg/m²) and half overweight to obese (BMI 28.0-39.9 kg/m²). Participants consumed two four-week controlled diets in randomized order: 1) a whole grain diet (WG diet), designed to be low in glycemic load (GL), high in whole grains, legumes, and fruits and vegetables, and 2) a refined grain diet (RG diet), designed to be high GL, high in refined grains and added sugars, separated by a four-week washout period. Quantitative targeted analysis of 55 bile acid species in fasting plasma was performed using liquid chromatography tandem mass spectrometry. Concentrations of glucose, insulin, and CRP were measured in fasting serum. Linear mixed models were used to test the effects of diet on bile acid concentrations, and determine the association between plasma bile acid concentrations and HOMA-IR and CRP. Benjamini-Hochberg false discovery rate (FDR) < 0.05 was used to control for multiple testing.

RESULTS

A total of 29 plasma bile acids were reliably detected and retained for analysis. Taurolithocholic acid (TLCA), taurocholic acid (TCA) and glycocholic acid (GCA) were statistically significantly higher after the WG compared to the RG diet (FDR < 0.05). There were no significant differences by BMI or sex. When evaluating the association of bile acids and HOMA-IR, GCA, taurochenodeoxycholic acid, ursodeoxycholic acid (UDCA), 5β‑cholanic acid‑3β,12α‑diol, 5‑cholanic acid‑3β‑ol, and glycodeoxycholic acid (GDCA) were statistically significantly positively associated with HOMA-IR individually, and as a group, total, 12α‑hydroxylated, primary and secondary bile acids were also significant (FDR < 0.05). When stratifying by BMI, chenodeoxycholic acid (CDCA), cholic acid (CA), UDCA, 5β-cholanic acid-3β, deoxycholic acid, and total, 12α-hydroxylated, primary and secondary bile acid groups were significantly positively associated with HOMA-IR among overweight to obese individuals (FDR < 0.05). When stratifying by sex, GCA, CDCA, TCA, CA, UDCA, GDCA, glycolithocholic acid (GLCA), total, primary, 12α‑hydroxylated, and glycine-conjugated bile acids were significantly associated with HOMA-IR among women, and CDCA, GDCA, and GLCA were significantly associated among men (FDR < 0.05). There were no significant associations between bile acids and CRP.

CONCLUSIONS

Diets with comparable macronutrient and energy composition, but differing in carbohydrate source, affected fasting plasma bile acids differently. Specifically, a diet characterized by whole grains, legumes, and fruits and vegetables compared to a diet high in refined grains and added sugars led to modest increases in concentrations of TLCA, TCA and GCA, ligands for FXR and TGR5, which may have beneficial effects on glucose homeostasis.

KERNEL-PENALIZED REGRESSION FOR ANALYSIS OF MICROBIOME DATA

The analysis of human microbiome data is often based on dimension-reduced graphical displays and clusterings derived from vectors of microbial abundances in each sample. Common to these ordination methods is the use of biologically motivated definitions of similarity. Principal coordinate analysis, in particular, is often performed using ecologically defined distances, allowing analyses to incorporate context-dependent, non-Euclidean structure. In this paper, we go beyond dimension-reduced ordination methods and describe a framework of high-dimensional regression models that extends these distance-based methods. In particular, we use kernel-based methods to show how to incorporate a variety of extrinsic information, such as phylogeny, into penalized regression models that estimate taxonspecific associations with a phenotype or clinical outcome. Further, we show how this regression framework can be used to address the compositional nature of multivariate predictors comprised of relative abundances; that is, vectors whose entries sum to a constant. We illustrate this approach with several simulations using data from two recent studies on gut and vaginal microbiomes. We conclude with an application to our own data, where we also incorporate a significance test for the estimated coefficients that represent associations between microbial abundance and a percent fat.

Plasma metabolite abundances are associated with urinary enterolactone excretion in healthy participants on controlled diets

Enterolignans, products of gut bacterial metabolism of plant lignans, have been associated with reduced risk of chronic diseases, but their association with other plasma metabolites is unknown. We examined plasma metabolite profiles according to urinary enterolignan excretion in a cross-sectional analysis using data from a randomized crossover, controlled feeding study. Eighty healthy adult males and females completed two 28-day feeding periods differing by glycemic load, refined carbohydrate, and fiber content. Lignan intake was calculated from food records using a polyphenol database. Targeted metabolomics was performed by LC-MS on plasma from fasting blood samples collected at the end of each feeding period. Enterolactone (ENL) and enterodiol, were measured in 24 h urine samples collected on the penultimate day of each study period using GC-MS. Linear mixed models were used to test the association between enterolignan excretion and metabolite abundances. Pathway analyses were conducted using the Global Test. Benjamini-Hochberg false discovery rate (FDR) was used to control for multiple testing. Of the metabolites assayed, 121 were detected in all samples. ENL excretion was associated positively with plasma hippuric acid and melatonin, and inversely with epinephrine, creatine, glycochenodeoxycholate, and glyceraldehyde (P < 0.05). Hippuric acid only satisfied the FDR of q < 0.1. END excretion was associated with myristic acid and glycine (q < 0.5). Two of 57 pathways tested were associated significantly with ENL, ubiquinone and terpenoid-quinone biosynthesis, and inositol phosphate metabolism. These results suggest a potential role for ENL or ENL-metabolizing gut bacteria in regulating plasma metabolites.

Nearly assumptionless screening for the mutually-exciting multivariate Hawkes process

We consider the task of learning the structure of the graph underlying a mutually-exciting multivariate Hawkes process in the high-dimensional setting. We propose a simple and computationally inexpensive edge screening approach. Under a subset of the assumptions required for penalized estimation approaches to recover the graph, this edge screening approach has the sure screening property: with high probability, the screened edge set is a superset of the true edge set. Furthermore, the screened edge set is relatively small. We illustrate the performance of this new edge screening approach in simulation studies.

Using Twitter for Demographic and Social Science Research: Tools for Data Collection and Processing

Despite recent and growing interest in using Twitter to examine human behavior and attitudes, there is still significant room for growth regarding the ability to leverage Twitter data for social science research. In particular, gleaning demographic information about Twitter users-a key component of much social science research-remains a challenge. This article develops an accurate and reliable data processing approach for social science researchers interested in using Twitter data to examine behaviors and attitudes, as well as the demographic characteristics of the populations expressing or engaging in them. Using information gathered from Twitter users who state an intention to not vote in the 2012 presidential election, we describe and evaluate a method for processing data to retrieve demographic information reported by users that is not encoded as text (e.g., details of images) and evaluate the reliability of these techniques. We end by assessing the challenges of this data collection strategy and discussing how large-scale social media data may benefit demographic researchers.

Exploratory plasma proteomic analysis in a randomized crossover trial of aspirin among healthy men and women

Long-term use of aspirin is associated with lower risk of colorectal cancer and other cancers; however, the mechanism of chemopreventive effect of aspirin is not fully understood. Animal studies suggest that COX-2, NFκB signaling and Wnt/β-catenin pathways may play a role, but no clinical trials have systematically evaluated the biological response to aspirin in healthy humans. Using a high-density antibody array, we assessed the difference in plasma protein levels after 60 days of regular dose aspirin (325 mg/day) compared to placebo in a randomized double-blinded crossover trial of 44 healthy non-smoking men and women, aged 21-45 years. The plasma proteome was analyzed on an antibody microarray with ~3,300 full-length antibodies, printed in triplicate. Moderated paired t-tests were performed on individual antibodies, and gene-set analyses were performed based on KEGG and GO pathways. Among the 3,000 antibodies analyzed, statistically significant differences in plasma protein levels were observed for nine antibodies after adjusting for false discoveries (FDR adjusted p-value<0.1). The most significant protein was succinate dehydrogenase subunit C (SDHC), a key enzyme complex of the mitochondrial tricarboxylic acid (TCA) cycle. The other statistically significant proteins (NR2F1, MSI1, MYH1, FOXO1, KHDRBS3, NFKBIE, LYZ and IKZF1) are involved in multiple pathways, including DNA base-pair repair, inflammation and oncogenic pathways. None of the 258 KEGG and 1,139 GO pathways was found to be statistically significant after FDR adjustment. This study suggests several chemopreventive mechanisms of aspirin in humans, which have previously been reported to play a role in anti- or pro-carcinogenesis in cell systems; however, larger, confirmatory studies are needed.

Transcriptional networks are associated with resistance to Mycobacterium tuberculosis infection

RATIONALE

Understanding mechanisms of resistance to M. tuberculosis (M.tb) infection in humans could identify novel therapeutic strategies as it has for other infectious diseases, such as HIV.

OBJECTIVES

To compare the early transcriptional response of M.tb-infected monocytes between Ugandan household contacts of tuberculosis patients who demonstrate clinical resistance to M.tb infection (cases) and matched controls with latent tuberculosis infection.

METHODS

Cases (n = 10) and controls (n = 18) were selected from a long-term household contact study in which cases did not convert their tuberculin skin test (TST) or develop tuberculosis over two years of follow up. We obtained genome-wide transcriptional profiles of M.tb-infected peripheral blood monocytes and used Gene Set Enrichment Analysis and interaction networks to identify cellular processes associated with resistance to clinical M.tb infection.

MEASUREMENTS AND MAIN RESULTS

We discovered gene sets associated with histone deacetylases that were differentially expressed when comparing resistant and susceptible subjects. We used small molecule inhibitors to demonstrate that histone deacetylase function is important for the pro-inflammatory response to in-vitro M.tb infection in human monocytes.

CONCLUSIONS

Monocytes from individuals who appear to resist clinical M.tb infection differentially activate pathways controlled by histone deacetylase in response to in-vitro M.tb infection when compared to those who are susceptible and develop latent tuberculosis. These data identify a potential cellular mechanism underlying the clinical phenomenon of resistance to M.tb infection despite known exposure to an infectious contact.

Network Reconstruction From High-Dimensional Ordinary Differential Equations

We consider the task of learning a dynamical system from high-dimensional time-course data. For instance, we might wish to estimate a gene regulatory network from gene expression data measured at discrete time points. We model the dynamical system nonparametrically as a system of additive ordinary differential equations. Most existing methods for parameter estimation in ordinary differential equations estimate the derivatives from noisy observations. This is known to be challenging and inefficient. We propose a novel approach that does not involve derivative estimation. We show that the proposed method can consistently recover the true network structure even in high dimensions, and we demonstrate empirical improvement over competing approaches. Supplementary materials for this article are available online.

Abstract 1056: Inhibition of hexose monophosphate pathway promotes castration resistant prostate cancer

Prostate Cancer (PCa) is the second highest cause of cancer-related death in men in the US. PCa is androgen dependent when organ-confined and is conventionally treated using surgery or using a combination of anti-androgens and radiation therapy. However, in about 30% of the patients tumor recurs and are initially administered androgen deprivation therapy (ADT). Majority of the patients become resistant to ADT and develop hormone-refractory disease also termed castration-resistant prostate cancer (CRPC), which is lethal. Currently, the molecular and biochemical alterations driving CRPC are not well understood. Using a novel network-based integrative approach, we show distinct alterations in the Hexosamine Biosynthetic Pathway (HBP) to be critical for sustaining the castrate resistant state. Our data suggests expression of key HBP enzymes to be significantly elevated in androgen dependent (AD) PCa while interestingly enough, relatively diminished in CRPC. Genetic loss of function experiments for these HBP enzymes in CRPC-like cells had tumor promoting effect both in vitro and in vivo. This was mediated by alterations in either PI3K-AKT pathway or SP1-ChREBP (SP1- carbohydrate response element binding protein) network in CRPC cells containing full length androgen receptor (AR) or its splice variant AR-V7, respectively. Strikingly, addition of HBP metabolite UDP-N-acetylglucosamine (UDP-GlcNAc) or glucosamine (GlcN) to CRPC-like cells attenuated tumor cell proliferation, both in vitro and in animal studies. Interestingly, these metabolites demonstrated additive efficacy when combined with enzalutamide in vitro. These findings are particularly significant given that the CRPC-like cells tested, inclusive of those containing AR-V7 variant, are inherently resistant to enzalutamide. These observations demonstrate the therapeutic value of targeting altered HBP in CRPC.

Citation Format: Akash K. Kaushik, Ali Shojaie, Katrin Panzitt, Rajni Sonavane, Harene Venghatakrishnan, Mohan Manikkam, Alexander Zaslavsky, Vasanta Putluri, Vihas Vasu, Yiqing Zhang, Ayesha Khan, Stacy Lloyd, Adam Szafran, Subhamoy Dasgupta, David Bader, Fabio Stossi, Hangwen Li, Susmita Samanta, Xuhong Cao, Efrosini Tsouko, Shixia Huang, Daniel Frigo, Lawrence Chan, Dean Edwards, Benny Kaipparettu, Nicholas Mitsiades, Nancy Weigel, Michael Mancini, Michael Ittmann, Arul Chinnaiyan, Nagireddy Putluri, Ganesh Palapattu, George Michailidis, Arun Sreekumar. Inhibition of hexose monophosphate pathway promotes castration resistant prostate cancer. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 1056.

Abstract 4284: Exploratory plasma proteomic analysis in a randomized cross-over trial of aspirin among healthy individuals

Background: Long-term use of aspirin is associated with lower colorectal cancer (CRC) incidence; however, the mechanism of the chemopreventive effect of aspirin is not fully understood. Animal studies suggest that COX-2, NFκB signaling and Wnt/β-catenin pathways may play a role, but no clinical trials have systematically evaluated the biological response to aspirin in healthy humans. Methods: We assessed the difference in plasma protein levels after 60 days of regular dose aspirin (325 mg/day) compared to placebo in a randomized, double-blinded, placebo-controlled, cross-over trial of 44 healthy non-smoking men and women, aged 21-45 years. Plasma proteomics was analyzed on an antibody microarray with ∼3,000 full-length antibodies, printed in triplicate. Moderated paired t-test was performed on individual antibodies, and gene set analyses were performed for KEGG and GO pathways. Results: Among the 3,387 antibodies, significant differences in plasma protein levels were observed for 267 antibodies (p&lt;0.05), the most significant protein being a transcription-factor regulator belonging to a steroid receptor family and found to be differentially expressed in colon cancer cells. Other significant proteins are involved in multiple oncogenic pathways related to colon tumorigenesis. In the pathway analysis, 4 KEGG (among 138) and 69 GO (among 1,089) pathways were found to be significant (p&lt;0.05), including natural killer (NK) cell-mediated cytotoxicity, butanoate metabolism and Wnt signaling pathways. Pathways that modulate cellular protein binding to steroid receptors were also significantly different between aspirin treatment and placebo (p&lt;0.05). None of the results remained statistically significant after correction for multiple testing. Conclusion: Several proteins and pathways, which have previously been reported as playing a role in colorectal carcinogenesis in vitro were found to be differentially expressed after aspirin treatment vs. placebo in healthy human subjects. This study suggests several chemopreventive mechanisms of aspirin; however, larger, confirmatory studies are needed.

Citation Format: Xiaoliang Wang, Yuzheng Zhang, Ali Shojaie, Paul D. Lampe, Lisa Levy, Ulrike Peters, John D. Potter, Emily White, Johanna W. Lampe. Exploratory plasma proteomic analysis in a randomized cross-over trial of aspirin among healthy individuals. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 4284.

Network-based pathway enrichment analysis with incomplete network information

MOTIVATION

Pathway enrichment analysis has become a key tool for biomedical researchers to gain insight into the underlying biology of differentially expressed genes, proteins and metabolites. It reduces complexity and provides a system-level view of changes in cellular activity in response to treatments and/or in disease states. Methods that use existing pathway network information have been shown to outperform simpler methods that only take into account pathway membership. However, despite significant progress in understanding the association amongst members of biological pathways, and expansion of data bases containing information about interactions of biomolecules, the existing network information may be incomplete or inaccurate and is not cell-type or disease condition-specific.

RESULTS

We propose a constrained network estimation framework that combines network estimation based on cell- and condition-specific high-dimensional Omics data with interaction information from existing data bases. The resulting pathway topology information is subsequently used to provide a framework for simultaneous testing of differences in expression levels of pathway members, as well as their interactions. We study the asymptotic properties of the proposed network estimator and the test for pathway enrichment, and investigate its small sample performance in simulated and real data settings.

AVAILABILITY AND IMPLEMENTATION

The proposed method has been implemented in the R-package netgsa available on CRAN.

CONTACT

jinma@upenn.eduSupplementary information: Supplementary data are available at Bioinformatics online.

Joint Estimation of Precision Matrices in Heterogeneous Populations

We introduce a general framework for estimation of inverse covariance, or precision, matrices from heterogeneous populations. The proposed framework uses a Laplacian shrinkage penalty to encourage similarity among estimates from disparate, but related, subpopulations, while allowing for differences among matrices. We propose an efficient alternating direction method of multipliers (ADMM) algorithm for parameter estimation, as well as its extension for faster computation in high dimensions by thresholding the empirical covariance matrix to identify the joint block diagonal structure in the estimated precision matrices. We establish both variable selection and norm consistency of the proposed estimator for distributions with exponential or polynomial tails. Further, to extend the applicability of the method to the settings with unknown populations structure, we propose a Laplacian penalty based on hierarchical clustering, and discuss conditions under which this data-driven choice results in consistent estimation of precision matrices in heterogenous populations. Extensive numerical studies and applications to gene expression data from subtypes of cancer with distinct clinical outcomes indicate the potential advantages of the proposed method over existing approaches.

Identifying functional cancer-specific miRNA-mRNA interactions in testicular germ cell tumor

Testicular cancer is the most common cancer in men aged between 15 and 35 and more than 90% of testicular neoplasms are originated at germ cells. Recent research has shown the impact of microRNAs (miRNAs) in different types of cancer, including testicular germ cell tumor (TGCT). MicroRNAs are small non-coding RNAs which affect the development and progression of cancer cells by binding to mRNAs and regulating their expressions. The identification of functional miRNA-mRNA interactions in cancers, i.e. those that alter the expression of genes in cancer cells, can help delineate post-regulatory mechanisms and may lead to new treatments to control the progression of cancer. A number of sequence-based methods have been developed to predict miRNA-mRNA interactions based on the complementarity of sequences. While necessary, sequence complementarity is, however, not sufficient for presence of functional interactions. Alternative methods have thus been developed to refine the sequence-based interactions using concurrent expression profiles of miRNAs and mRNAs. This study aims to find functional cancer-specific miRNA-mRNA interactions in TGCT. To this end, the sequence-based predicted interactions are first refined using an ensemble learning method, based on two well-known methods of learning miRNA-mRNA interactions, namely, TaLasso and GenMiR++. Additional functional analyses were then used to identify a subset of interactions to be most likely functional and specific to TGCT. The final list of 13 miRNA-mRNA interactions can be potential targets for identifying TGCT-specific interactions and future laboratory experiments to develop new therapies.

Estimation of High-Dimensional Graphical Models Using Regularized Score Matching

Graphical models are widely used to model stochastic dependences among large collections of variables. We introduce a new method of estimating undirected conditional independence graphs based on the score matching loss, introduced by Hyvärinen (2005), and subsequently extended in Hyvärinen (2007). The regularized score matching method we propose applies to settings with continuous observations and allows for computationally efficient treatment of possibly non-Gaussian exponential family models. In the well-explored Gaussian setting, regularized score matching avoids issues of asymmetry that arise when applying the technique of neighborhood selection, and compared to existing methods that directly yield symmetric estimates, the score matching approach has the advantage that the considered loss is quadratic and gives piecewise linear solution paths under ℓ₁ regularization. Under suitable irrepresentability conditions, we show that ℓ₁-regularized score matching is consistent for graph estimation in sparse high-dimensional settings. Through numerical experiments and an application to RNAseq data, we confirm that regularized score matching achieves state-of-the-art performance in the Gaussian case and provides a valuable tool for computationally efficient estimation in non-Gaussian graphical models.

A significance test for graph-constrained estimation

Graph-constrained estimation methods encourage similarities among neighboring covariates presented as nodes of a graph, and can result in more accurate estimates, especially in high-dimensional settings. Variable selection approaches can then be utilized to select a subset of variables that are associated with the response. However, existing procedures do not provide measures of uncertainty of estimates. Further, the vast majority of existing approaches assume that available graph accurately captures the association among covariates; violations to this assumption could severely hurt the reliability of the resulting estimates. In this article, we present a new inference framework, called the Grace test, which produces coefficient estimates and corresponding p-values by incorporating the external graph information. We show, both theoretically and via numerical studies, that the proposed method asymptotically controls the type-I error rate regardless of the choice of the graph. We also show that when the underlying graph is informative, the Grace test is asymptotically more powerful than similar tests that ignore the external information. We study the power properties of the proposed test when the graph is not fully informative and develop a more powerful Grace-ridge test for such settings. Our numerical studies show that as long as the graph is reasonably informative, the proposed inference procedures deliver improved statistical power over existing methods that ignore external information.

The cluster graphical lasso for improved estimation of Gaussian graphical models

The task of estimating a Gaussian graphical model in the high-dimensional setting is considered. The graphical lasso, which involves maximizing the Gaussian log likelihood subject to a lasso penalty, is a well-studied approach for this task. A surprising connection between the graphical lasso and hierarchical clustering is introduced: the graphical lasso in effect performs a two-step procedure, in which (1) single linkage hierarchical clustering is performed on the variables in order to identify connected components, and then (2) a penalized log likelihood is maximized on the subset of variables within each connected component. Thus, the graphical lasso determines the connected components of the estimated network via single linkage clustering. The single linkage clustering is known to perform poorly in certain finite-sample settings. Therefore, the cluster graphical lasso, which involves clustering the features using an alternative to single linkage clustering, and then performing the graphical lasso on the subset of variables within each cluster, is proposed. Model selection consistency for this technique is established, and its improved performance relative to the graphical lasso is demonstrated in a simulation study, as well as in applications to a university webpage and a gene expression data sets.

Phenotypic and Molecular Characterization of Plasmid Mediated AmpC among Clinical Isolates of Klebsiella pneumoniae Isolated from Different Hospitals in Tehran

INTRODUCTION

Klebsiella pneumoniae is one of the main opportunistic pathogens which can cause different types of infections. Production of beta-lactamases like AmpC and ESBL mostly lead to beta-lactam resistance in these Gram-Negative bacteria. The aim of this study was the detection of AmpC-producing K. pneumoniae in clinical isolates.

MATERIALS AND METHODS

Three hundred and three isolates of K. pneumoniae were identified. Double disc method including cefoxitin with cefepime and using boronic acid with cloxacillin were performed as two phenotypic methods for detection of AmpC. Amplification of AmpC gene was performed by PCR.

RESULTS

Eight and three isolates showed positive results in double disc method and by using boronic acid with cloxacillin, respectively. Five isolates had specific band for AmpC gene after electrophoresis.

CONCLUSION

Our results were indicated the low prevalence of AmpC-producer-K. pnemoniae in Iran. On the other hand these two tested phenotypic methods showed low sensitivity for detection of AmpC.

Serial plasma metabolites following hypoxic-ischemic encephalopathy in a nonhuman primate model

Biomarkers that indicate the severity of hypoxic-ischemic brain injury and response to treatment and that predict neurodevelopmental outcomes are urgently needed to improve the care of affected neonates. We hypothesize that sequentially obtained plasma metabolomes will provide indicators of brain injury and repair, allowing for the prediction of neurodevelopmental outcomes. A total of 33 Macaca nemestrina underwent 0, 15 or 18 min of in utero umbilical cord occlusion (UCO) to induce hypoxic-ischemic encephalopathy and were then delivered by hysterotomy, resuscitated and stabilized. Serial blood samples were obtained at baseline (cord blood) and at 0.1, 24, 48, and 72 h of age. Treatment groups included nonasphyxiated controls (n = 7), untreated UCO (n = 11), UCO + hypothermia (HT; n = 6), and UCO + HT + erythropoietin (n = 9). Metabolites were extracted and analyzed using comprehensive two-dimensional gas chromatography coupled with time-of-flight mass spectrometry and quantified by PARAFAC (parallel factor analysis). Using nontargeted discovery-based methods, we identified 63 metabolites as potential biomarkers. The changes in metabolite concentrations were characterized and compared between treatment groups. Further comparison determined that 8 metabolites (arachidonic acid, butanoic acid, citric acid, fumaric acid, lactate, malate, propanoic acid, and succinic acid) correlated with early and/or long-term neurodevelopmental outcomes. The combined outcomes of death or cerebral palsy correlated with citric acid, fumaric acid, lactate, and propanoic acid. This change in circulating metabolome after UCO may reflect cellular metabolism and biochemical changes in response to the severity of brain injury and have potential to predict neurodevelopmental outcomes.

Network Granger Causality with Inherent Grouping Structure

The problem of estimating high-dimensional network models arises naturally in the analysis of many biological and socio-economic systems. In this work, we aim to learn a network structure from temporal panel data, employing the framework of Granger causal models under the assumptions of sparsity of its edges and inherent grouping structure among its nodes. To that end, we introduce a group lasso regression regularization framework, and also examine a thresholded variant to address the issue of group misspecification. Further, the norm consistency and variable selection consistency of the estimates are established, the latter under the novel concept of direction consistency. The performance of the proposed methodology is assessed through an extensive set of simulation studies and comparisons with existing techniques. The study is illustrated on two motivating examples coming from functional genomics and financial econometrics.

Selection and estimation for mixed graphical models

We consider the problem of estimating the parameters in a pairwise graphical model in which the distribution of each node, conditioned on the others, may have a different exponential family form. We identify restrictions on the parameter space required for the existence of a well-defined joint density, and establish the consistency of the neighbourhood selection approach for graph reconstruction in high dimensions when the true underlying graph is sparse. Motivated by our theoretical results, we investigate the selection of edges between nodes whose conditional distributions take different parametric forms, and show that efficiency can be gained if edge estimates obtained from the regressions of particular nodes are used to reconstruct the graph. These results are illustrated with examples of Gaussian, Bernoulli, Poisson and exponential distributions. Our theoretical findings are corroborated by evidence from simulation studies.

Comparative study of computational methods for reconstructing genetic networks of cancer-related pathways

Network reconstruction is an important yet challenging task in systems biology. While many methods have been recently proposed for reconstructing biological networks from diverse data types, properties of estimated networks and differences between reconstruction methods are not well understood. In this paper, we conduct a comprehensive empirical evaluation of seven existing network reconstruction methods, by comparing the estimated networks with different sparsity levels for both normal and tumor samples. The results suggest substantial heterogeneity in networks reconstructed using different reconstruction methods. Our findings also provide evidence for significant differences between networks of normal and tumor samples, even after accounting for the considerable variability in structures of networks estimated using different reconstruction methods. These differences can offer new insight into changes in mechanisms of genetic interaction associated with cancer initiation and progression.

The Cluster Elastic Net for High-Dimensional Regression With Unknown Variable Grouping

In the high-dimensional regression setting, the elastic net produces a parsimonious model by shrinking all coefficients towards the origin. However, in certain settings, this behavior might not be desirable: if some features are highly correlated with each other and associated with the response, then we might wish to perform less shrinkage on the coefficients corresponding to that subset of features. We propose the cluster elastic net, which selectively shrinks the coefficients for such variables towards each other, rather than towards the origin. Instead of assuming that the clusters are known a priori, the cluster elastic net infers clusters of features from the data, on the basis of correlation among the variables as well as association with the response. These clusters are then used in order to more accurately perform regression. We demonstrate the theoretical advantages of our proposed approach, and explore its performance in a simulation study, and in an application to HIV drug resistance data. Supplementary Materials are available online.

Metabolomic profiling identifies biochemical pathways associated with castration-resistant prostate cancer

Despite recent developments in treatment strategies, castration-resistant prostate cancer (CRPC) is still the second leading cause of cancer-associated mortality among American men, the biological underpinnings of which are not well understood. To this end, we measured levels of 150 metabolites and examined the rate of utilization of 184 metabolites in metastatic androgen-dependent prostate cancer (AD) and CRPC cell lines using a combination of targeted mass spectrometry and metabolic phenotyping. Metabolic data were used to derive biochemical pathways that were enriched in CRPC, using Oncomine concept maps (OCM). The enriched pathways were then examined in-silico for their association with treatment failure (i.e., prostate specific antigen (PSA) recurrence or biochemical recurrence) using published clinically annotated gene expression data sets. Our results indicate that a total of 19 metabolites were altered in CRPC compared to AD cell lines. These altered metabolites mapped to a highly interconnected network of biochemical pathways that describe UDP glucuronosyltransferase (UGT) activity. We observed an association with time to treatment failure in an analysis employing genes restricted to this pathway in three independent gene expression data sets. In summary, our studies highlight the value of employing metabolomic strategies in cell lines to derive potentially clinically useful predictive tools.

Graph Estimation with Joint Additive Models

In recent years, there has been considerable interest in estimating conditional independence graphs in high dimensions. Most previous work has assumed that the variables are multivariate Gaussian, or that the conditional means of the variables are linear; in fact, these two assumptions are nearly equivalent. Unfortunately, if these assumptions are violated, the resulting conditional independence estimates can be inaccurate. We propose a semi-parametric method, graph estimation with joint additive models, which allows the conditional means of the features to take on an arbitrary additive form. We present an efficient algorithm for our estimator's computation, and prove that it is consistent. We extend our method to estimation of directed graphs with known causal ordering. Using simulated data, we show that our method performs better than existing methods when there are non-linear relationships among the features, and is comparable to methods that assume multivariate normality when the conditional means are linear. We illustrate our method on a cell-signaling data set.

Using random walks to identify cancer-associated modules in expression data

BACKGROUND

The etiology of cancer involves a complex series of genetic and environmental conditions. To better represent and study the intricate genetics of cancer onset and progression, we construct a network of biological interactions to search for groups of genes that compose cancer-related modules. Three cancer expression datasets are investigated to prioritize genes and interactions associated with cancer outcomes. Using a graph-based approach to search for communities of phenotype-related genes in microarray data, we find modules of genes associated with cancer phenotypes in a weighted interaction network.

RESULTS

We implement Walktrap, a random-walk-based community detection algorithm, to identify biological modules predisposing to tumor growth in 22 hepatocellular carcinoma samples (GSE14520), adenoma development in 32 colorectal cancer samples (GSE8671), and prognosis in 198 breast cancer patients (GSE7390). For each study, we find the best scoring partitions under a maximum cluster size of 200 nodes. Significant modules highlight groups of genes that are functionally related to cancer and show promise as therapeutic targets; these include interactions among transcription factors (SPIB, RPS6KA2 and RPS6KA6), cell-cycle regulatory genes (BRSK1, WEE1 and CDC25C), modulators of the cell-cycle and proliferation (CBLC and IRS2) and genes that regulate and participate in the map-kinase pathway (MAPK9, DUSP1, DUSP9, RIPK2). To assess the performance of Walktrap to find genomic modules (Walktrap-GM), we evaluate our results against other tools recently developed to discover disease modules in biological networks. Compared with other highly cited module-finding tools, jActiveModules and Matisse, Walktrap-GM shows strong performance in the discovery of modules enriched with known cancer genes.

CONCLUSIONS

These results demonstrate that the Walktrap-GM algorithm identifies modules significantly enriched with cancer genes, their joint effects and promising candidate genes. The approach performs well when evaluated against similar tools and smaller overall module size allows for more specific functional annotation and facilitates the interpretation of these modules.

Abstract 5387: Integrative analysis of transcriptomic and metabolomic data reveals a critical role for aminosugar metabolism in prostate cancer

Prostate Cancer is the second most common cause of cancer-related death in men in the US. Like other tumors, prostate cancer development and progression is dictated by multiple molecular events that include changes in levels of genes, transcripts, proteins and metabolites. To better understand the biology of prostate cancer development it is essential to integrate these disparate yet related datasets. Our laboratory has identified matched transcriptomic, proteomic and metabolomic changes in localized prostate cancer relative to adjacent benign tissue as well in metastatic disease compared to organ-confined tumor. To study this using a System's Biology approach we have recently embarked on a pilot study aimed at integrating transcriptomic and metabolomic data to obtain biochemical pathways that key to prostate cancer development. Using an in-house network-based enrichment strategy, amino sugar metabolism was found to be significantly enriched in organ-confined prostate cancer but not in metastatic disease. Amino sugar metabolism describes the utilization of glucose-derived carbon and amino acid (mostly glutamine)-derive nitrogen to produce glucosamines. These amino sugars participate in synthesis of immune modulatory compounds as well as in glycosylation cascades. In this study, we describe the molecular analyses of Glucosamine-6 phosphate-N-acetyl Transferase (GNPNAT1), a key enzyme that converts D-glucosamine 6-phosphate to N-acetyl-D-glucosamine 6-phosphate, in prostate cancer. Our results indicate upregulation of GNPNAT1 in organ-confined prostate cancer as compared to benign adjacent tissue as well as metastatic tissue and regulation of the pathway by androgen. Stable knockdown of GNPNAT1 in androgen dependent LNCap cells leads to diminished cell growth and cell cycle arrest. In contrast, growth and cell cycle are not affected by the knockdown of GNPNAT1 in androgen independent C4-2 cells. Knockdown of GNPNAT1 in C4-2 cells enhances invasiveness which is not observed in LNCap knockdown cells. In this work we show that GNPNAT1 is linked to androgen receptor (AR) action and thus is a critical enzyme for the survival of androgen dependent prostate cancer cells.

Citation Format: Katrin Panzitt, Ali Shojaie, Nagireddy Putluri, Sumanta Basu, Vasanta Putluri, Susmita Samanta, Michael Ittmann, Ismael Vergara, George Michailidis, Ganesh Palapattu, Arun Sreekumar. Integrative analysis of transcriptomic and metabolomic data reveals a critical role for aminosugar metabolism in prostate cancer. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 5387. doi:10.1158/1538-7445.AM2013-5387

Metabolomic profiling reveals potential markers and bioprocesses altered in bladder cancer progression

Although alterations in xenobiotic metabolism are considered causal in the development of bladder cancer, the precise mechanisms involved are poorly understood. In this study, we used high-throughput mass spectrometry to measure over 2,000 compounds in 58 clinical specimens, identifying 35 metabolites which exhibited significant changes in bladder cancer. This metabolic signature distinguished both normal and benign bladder from bladder cancer. Exploratory analyses of this metabolomic signature in urine showed promise in distinguishing bladder cancer from controls and also nonmuscle from muscle-invasive bladder cancer. Subsequent enrichment-based bioprocess mapping revealed alterations in phase I/II metabolism and suggested a possible role for DNA methylation in perturbing xenobiotic metabolism in bladder cancer. In particular, we validated tumor-associated hypermethylation in the cytochrome P450 1A1 (CYP1A1) and cytochrome P450 1B1 (CYP1B1) promoters of bladder cancer tissues by bisulfite sequence analysis and methylation-specific PCR and also by in vitro treatment of T-24 bladder cancer cell line with the DNA demethylating agent 5-aza-2'-deoxycytidine. Furthermore, we showed that expression of CYP1A1 and CYP1B1 was reduced significantly in an independent cohort of bladder cancer specimens compared with matched benign adjacent tissues. In summary, our findings identified candidate diagnostic and prognostic markers and highlighted mechanisms associated with the silencing of xenobiotic metabolism. The metabolomic signature we describe offers potential as a urinary biomarker for early detection and staging of bladder cancer, highlighting the utility of evaluating metabolomic profiles of cancer to gain insights into bioprocesses perturbed during tumor development and progression.

Metabolomic profiling reveals a role for androgen in activating amino acid metabolism and methylation in prostate cancer cells

Prostate cancer is the second leading cause of cancer related death in American men. Development and progression of clinically localized prostate cancer is highly dependent on androgen signaling. Metastatic tumors are initially responsive to anti-androgen therapy, however become resistant to this regimen upon progression. Genomic and proteomic studies have implicated a role for androgen in regulating metabolic processes in prostate cancer. However, there have been no metabolomic profiling studies conducted thus far that have examined androgen-regulated biochemical processes in prostate cancer. Here, we have used unbiased metabolomic profiling coupled with enrichment-based bioprocess mapping to obtain insights into the biochemical alterations mediated by androgen in prostate cancer cell lines. Our findings indicate that androgen exposure results in elevation of amino acid metabolism and alteration of methylation potential in prostate cancer cells. Further, metabolic phenotyping studies confirm higher flux through pathways associated with amino acid metabolism in prostate cancer cells treated with androgen. These findings provide insight into the potential biochemical processes regulated by androgen signaling in prostate cancer. Clinically, if validated, these pathways could be exploited to develop therapeutic strategies that supplement current androgen ablative treatments while the observed androgen-regulated metabolic signatures could be employed as biomarkers that presage the development of castrate-resistant prostate cancer.

Use of metabolomic profiling to identify potential markers and mechanism for bladder cancer progression

250

Background: Bladder cancer (BCa) is the second most prevalent urological malignancy and the fourth highest cause of cancer-related death in the United States that is known to be caused by defects in xenobiotic metabolism, which is not well understood. Using unbiased mass spectrometry we report the metabolomic profiles in BCa tissues and demonstrate its biomarker potential as well reveal a possible mechanism regulating altered xenobiotic metabolism in these tumors.

Methods: Total metabolome from clinically annotated bladder-derived tissues were examined using mass spectrometry coupled to reverse and aqueous normal phase separation of compounds. Class- specific metabolites were examined in urine specimens for their biomarker potential as well as analyzed using Oncomine Concept Map for alterations in bioprocesses. The latter was validated using a collection of molecular techniques like Q-PCR, immunoblot analysis, methylation assays and use of methyl transferase inhibitor, on bladder-derived cell lines and patients specimens.

Results: A total of 2,019 compounds were detected across the 58 bladder-derived specimens, of which 50 named compounds were differential between BCa and its adjacent benign tissue. These included aromatic compounds like aniline, catechols as well as polyamines and S-adenosyl methionine (SAM). A subset of these compounds, were detected in urine and could distinguish BCa from benign, non-muscle-invasive from muscle invasive tumors and delineate patients responding to chemotherapy and TURBT. Bioprocess mapping of BCa-specific metabolome revealed co-enriched concepts describing methylation and cytochrome P450 (CYP) driven xenobiotic metabolism in bladder tumors. The role of methylation in regulation of CYP activity in BCa was confirmed using a combination of Aza-treatment, methylation-specific PCR and bisulphite sequencing on bladder-derived cell lines and tissues.

Conclusions: Unbiased metabolomic profiling reveals potential non-invasive metabolic markers for early detection, prognosis and therapeutic response of BCa as well as describes a role for methylation induced silencing of CYP1A1 and 1B1resulting in deficient xenobiotic metabolism in BCa.

No significant financial relationships to disclose.

Discovering graphical Granger causality using the truncating lasso penalty

MOTIVATION

Components of biological systems interact with each other in order to carry out vital cell functions. Such information can be used to improve estimation and inference, and to obtain better insights into the underlying cellular mechanisms. Discovering regulatory interactions among genes is therefore an important problem in systems biology. Whole-genome expression data over time provides an opportunity to determine how the expression levels of genes are affected by changes in transcription levels of other genes, and can therefore be used to discover regulatory interactions among genes.

RESULTS

In this article, we propose a novel penalization method, called truncating lasso, for estimation of causal relationships from time-course gene expression data. The proposed penalty can correctly determine the order of the underlying time series, and improves the performance of the lasso-type estimators. Moreover, the resulting estimate provides information on the time lag between activation of transcription factors and their effects on regulated genes. We provide an efficient algorithm for estimation of model parameters, and show that the proposed method can consistently discover causal relationships in the large p, small n setting. The performance of the proposed model is evaluated favorably in simulated, as well as real, data examples.

AVAILABILITY

The proposed truncating lasso method is implemented in the R-package 'grangerTlasso' and is freely available at http://www.stat.lsa.umich.edu/~shojaie/.

Abstract A52: Metabolomic profiling reveals impaired xenobiotic metabolism in bladder cancer

Introduction: Bladder cancer (BCa) is the second most prevalent urological malignancy and the fourth highest cause of cancer-related death in the United States. Earlier studies have linked BCa development to alterations in metabolic pathways. Significant among these are decreased activity of N-acetyl transferases causing a slow-acetylator phenotype leading to inefficient detoxification of aromatic hydrocarbons causal to onset of BCa. Interestingly, Afro-American patients inherently exhibit such slow acetylator phenotype and are known to have a more aggressive form of the tumor compared to Caucasians. This indicates existence of a metabolic niche that governs the racial disparity in BCa, which is not well understood. Also, there is an imminent need to develop non-invasive markers for early detection and prognosis of BCa, since urine cytology which is the current clinical standard is not specific to the tumor. Using mass spectrometry we report metabolic alterations in BCa and delineate bioprocesses that are altered during its progression. Our data for the first-time demonstrates role of methylation in attenuation of xenotbiotic metabolism in BCa. Furthermore, the metabolic profiles seed further analysis to examine the racial disparity in these tumors.

Methods: Total metabolome from flash frozen clinically annotated bladder-derived tissues (n=58,31 benign adjacent and 27 BCa, 26 matched pairs) were examined using a combination of Q-TOF (unbiased) and triple-quadrupole (targeted) mass spectrometry. Panel of well-defined standards were used to ensure reproducibility of the profiling process. The metabolites were pre-fractionated using liquid chromatography prior to mass spectrometry in both the positive and negative ionization mode. The unbiased mass spectral data was searched using Metlin library to identify the compounds. The metabolomic profiles thus generated were analyzed to delineate class-specific signatures which were interrogated for altered bioprocesses using Molecular Concept Map (OCM, www.oncomine.org). The altered bioprocesses were validated in cell line models using a combination of Q-PCR, immunoblot analysis and functional assays.

Results and discussion: A total of 2019 compounds were detected across the 58 bladder-derived specimens of which, 423 compounds were significantly altered in BCa compared to adjacent benign. 50 of the differential compounds were named and used for developing a classificatory signature and bioprocess mapping. Included among these were polycylic compounds like aniline, catechols, aromatic amino acids, polyamines and S-adenosyl methionine (SAM). Interestingly this BCa-specific metabolic signature in tissues was able to delineate tumor from benign with an accuracy of 75 %. Importantly the functional mapping of the metabolic data revealed enhanced methylation potential in tumors as being one of the factors de-regulating the xenobiotic metabolism. In vitro experiments using bisulfite sequencing and methyltransferase inhibitor 5-Aza-cytidine confirmed this methylation-induced attenuation of phase I/II metabolic genes namely CYP1A1, CYP1B1, EPHX1 and GSTT1 in BCa. In summary, using unbiased metabolomic profiling report metabolic fingerprint for bladder cancer. Importantly our data for the first time reveals methylation-induced silencing of xenobiotic metabolism in bladder tumors.

Citation Information: Cancer Epidemiol Biomarkers Prev 2010;19(10 Suppl):A52.

Penalized likelihood methods for estimation of sparse high-dimensional directed acyclic graphs

Directed acyclic graphs are commonly used to represent causal relationships among random variables in graphical models. Applications of these models arise in the study of physical and biological systems where directed edges between nodes represent the influence of components of the system on each other. Estimation of directed graphs from observational data is computationally NP-hard. In addition, directed graphs with the same structure may be indistinguishable based on observations alone. When the nodes exhibit a natural ordering, the problem of estimating directed graphs reduces to the problem of estimating the structure of the network. In this paper, we propose an efficient penalized likelihood method for estimation of the adjacency matrix of directed acyclic graphs, when variables inherit a natural ordering. We study variable selection consistency of lasso and adaptive lasso penalties in high-dimensional sparse settings, and propose an error-based choice for selecting the tuning parameter. We show that although the lasso is only variable selection consistent under stringent conditions, the adaptive lasso can consistently estimate the true graph under the usual regularity assumptions.

Analysis of gene sets based on the underlying regulatory network

Networks are often used to represent the interactions among genes and proteins. These interactions are known to play an important role in vital cell functions and should be included in the analysis of genes that are differentially expressed. Methods of gene set analysis take advantage of external biological information and analyze a priori defined sets of genes. These methods can potentially preserve the correlation among genes; however, they do not directly incorporate the information about the gene network. In this paper, we propose a latent variable model that directly incorporates the network information. We then use the theory of mixed linear models to present a general inference framework for the problem of testing the significance of subnetworks. Several possible test procedures are introduced and a network based method for testing the changes in expression levels of genes as well as the structure of the network is presented. The performance of the proposed method is compared with methods of gene set analysis using both simulation studies, as well as real data on genes related to the galactose utilization pathway in yeast.