Kourosh Zarringhalam

Integrating chemical footprinting data into RNA secondary structure prediction.

Chemical and enzymatic footprinting experiments, such as shape (selective 2′-hydroxyl acylation analyzed by primer extension), yield important information about RNA secondary structure. Indeed, since the -hydroxyl is reactive at flexible (loop) regions, but unreactive at base-paired regions, shape yields quantitative data about which RNA nucleotides are base-paired. Recently, low error rates in secondary structure prediction have been reported for three RNAs of moderate size, by including base stacking pseudo-energy terms derived from shape data into the computation of minimum free energy secondary structure. Here, we describe a novel method, RNAsc (RNA soft constraints), which includes pseudo-energy terms for each nucleotide position, rather than only for base stacking positions. We prove that RNAsc is self-consistent, in the sense that the nucleotide-specific probabilities of being unpaired in the low energy Boltzmann ensemble always become more closely correlated with the input shape data after application of RNAsc. From this mathematical perspective, the secondary structure predicted by RNAsc should be ‘correct’, in as much as the shape data is ‘correct’. We benchmark RNAsc against the previously mentioned method for eight RNAs, for which both shape data and native structures are known, to find the same accuracy in 7 out of 8 cases, and an improvement of 25% in one case. Furthermore, we present what appears to be the first direct comparison of shape data and in-line probing data, by comparing yeast asp-tRNA shape data from the literature with data from in-line probing experiments we have recently performed. With respect to several criteria, we find that shape data appear to be more robust than in-line probing data, at least in the case of asp-tRNA.

Statistical Analysis of the Processes Controlling Choline and Ethanolamine Glycerophospholipid Molecular Species Composition

The regulation and maintenance of the cellular lipidome through biosynthetic, remodeling, and catabolic mechanisms are critical for biological homeostasis during development, health and disease. These complex mechanisms control the architectures of lipid molecular species, which have diverse yet highly regulated fatty acid chains at both the sn1 and sn2 positions. Phosphatidylcholine (PC) and phosphatidylethanolamine (PE) serve as the predominant biophysical scaffolds in membranes, acting as reservoirs for potent lipid signals and regulating numerous enzymatic processes. Here we report the first rigorous computational dissection of the mechanisms influencing PC and PE molecular architectures from high-throughput shotgun lipidomic data. Using novel statistical approaches, we have analyzed multidimensional mass spectrometry-based shotgun lipidomic data from developmental mouse heart and mature mouse heart, lung, brain, and liver tissues. We show that in PC and PE, sn1 and sn2 positions are largely independent, though for low abundance species regulatory processes may interact with both the sn1 and sn2 chain simultaneously, leading to cooperative effects. Chains with similar biochemical properties appear to be remodeled similarly. We also see that sn2 positions are more regulated than sn1, and that PC exhibits stronger cooperative effects than PE. A key aspect of our work is a novel statistically rigorous approach to determine cooperativity based on a modified Fishers exact test using Markov Chain Monte Carlo sampling. This computational approach provides a novel tool for developing mechanistic insight into lipidomic regulation.

Molecular causes of transcriptional response: a Bayesian prior knowledge approach

MOTIVATION The abundance of many transcripts changes significantly in response to a variety of molecular and environmental perturbations. A key question in this setting is as follows: what intermediate molecular perturbations gave rise to the observed transcriptional changes? Regulatory programs are not exclusively governed by transcriptional changes but also by protein abundance and post-translational modifications making direct causal inference from data difficult. However, biomedical research over the last decades has uncovered a plethora of causal signaling cascades that can be used to identify good candidates explaining a specific set of transcriptional changes. METHODS We take a Bayesian approach to integrate gene expression profiling with a causal graph of molecular interactions constructed from prior biological knowledge. In addition, we define the biological context of a specific interaction by the corresponding Medical Subject Headings terms. The Bayesian network can be queried to suggest upstream regulators that can be causally linked to the altered expression profile. RESULTS Our approach will treat candidate regulators in the right biological context preferentially, enables hierarchical exploration of resulting hypotheses and takes the complete network of causal relationships into account to arrive at the best set of upstream regulators. We demonstrate the power of our method on distinct biological datasets, namely response to dexamethasone treatment, stem cell differentiation and a neuropathic pain model. In all cases relevant biological insights could be validated. AVAILABILITY AND IMPLEMENTATION Source code for the method is available upon request.

Dynamics of the ethanolamine glycerophospholipid remodeling network.

Acyl chain remodeling in lipids is a critical biochemical process that plays a central role in disease. However, remodeling remains poorly understood, despite massive increases in lipidomic data. In this work, we determine the dynamic network of ethanolamine glycerophospholipid (PE) remodeling, using data from pulse-chase experiments and a novel bioinformatic network inference approach. The model uses a set of ordinary differential equations based on the assumptions that (1) sn1 and sn2 acyl positions are independently remodeled; (2) remodeling reaction rates are constant over time; and (3) acyl donor concentrations are constant. We use a novel fast and accurate two-step algorithm to automatically infer model parameters and their values. This is the first such method applicable to dynamic phospholipid lipidomic data. Our inference procedure closely fits experimental measurements and shows strong cross-validation across six independent experiments with distinct deuterium-labeled PE precursors, demonstrating the validity of our assumptions. In constrast, fits of randomized data or fits using random model parameters are worse. A key outcome is that we are able to robustly distinguish deacylation and reacylation kinetics of individual acyl chain types at the sn1 and sn2 positions, explaining the established prevalence of saturated and unsaturated chains in the respective positions. The present study thus demonstrates that dynamic acyl chain remodeling processes can be reliably determined from dynamic lipidomic data.

Pressure dependency of localization degree in heavy fermion CeIn3: A density functional theory analysis.

Two dramatic discrepancies between previous reliable experimental and ab initio DFT results are identified to occur at two different pressures in CeIn3, as discussed through the paper. We physically discuss sources of the phenomena and indicate how to select an appropriate functional for a given pressure. We show that these discrepancies are due to the inaccuracy of the DFT + U scheme with arbitrary Ueff and that hybrid functionals can provide better agreement with experimental data at zero pressure. The hybrid B3PW91 approach provides much better agreement with experimental data than the GGA + U. The DFT + U scheme proves to be rather unreliable since it yields completely unpredictable oscillations for the bulk modulus with increasing values of Ueff. Our B3PW91 results show that the best lattice parameter (bulk modulus) is obtained using a larger value of α parameter, 0.4 (0.3 or 0.2), than that of usually considered for the AFM phase. We find that for hybrid functionals, the amount of non-local exchange must first be calibrated before conclusions are drawn. Therefore, we first systematically optimize the α parameter and using it investigate the magnetic and electronic properties of the system. We present a theoretical interpretation of the experimental results and reproduce them satisfactorily.

Interpreting transcriptional changes using causal graphs: new methods and their practical utility on public networks

BackgroundInference of active regulatory cascades under specific molecular and environmental perturbations is a recurring task in transcriptional data analysis. Commercial tools based on large, manually curated networks of causal relationships offering such functionality have been used in thousands of articles in the biomedical literature. The adoption and extension of such methods in the academic community has been hampered by the lack of freely available, efficient algorithms and an accompanying demonstration of their applicability using current public networks.ResultsIn this article, we propose a new statistical method that will infer likely upstream regulators based on observed patterns of up- and down-regulated transcripts. The method is suitable for use with public interaction networks with a mix of signed and unsigned causal edges. It subsumes and extends two previously published approaches and we provide a novel algorithmic method for efficient statistical inference. Notably, we demonstrate the feasibility of using the approach to generate biological insights given current public networks in the context of controlled in-vitro overexpression experiments, stem-cell differentiation data and animal disease models. We also provide an efficient implementation of our method in the R package QuaternaryProd available to download from Bioconductor.ConclusionsIn this work, we have closed an important gap in utilizing causal networks to analyze differentially expressed genes. Our proposed Quaternary test statistic incorporates all available evidence on the potential relevance of an upstream regulator. The new approach broadens the use of these types of statistics for highly curated signed networks in which ambiguities arise but also enables the use of networks with unsigned edges. We design and implement a novel computational method that can efficiently estimate p-values for upstream regulators in current biological settings. We demonstrate the ready applicability of the implemented method to analyze differentially expressed genes using the publicly available networks.

Identification of competing endogenous RNAs of the tumor suppressor gene PTEN: A probabilistic approach

Regulation by microRNAs (miRNAs) and modulation of miRNA activity are critical components of diverse cellular processes. Recent research has shown that miRNA-based regulation of the tumor suppressor gene PTEN can be modulated by the expression of other miRNA targets acting as competing endogenous RNAs (ceRNAs). However, the key sequence-based features enabling a transcript to act as an effective ceRNA are not well understood and a quantitative model associating statistical significance to such features is currently lacking. To identify and assess features characterizing target recognition by PTEN-regulating miRNAs, we analyze multiple datasets from PAR-CLIP experiments in conjunction with RNA-Seq data. We consider a set of miRNAs known to regulate PTEN and identify high-confidence binding sites for these miRNAs on the 3′ UTR of protein coding genes. Based on the number and spatial distribution of these binding sites, we calculate a set of probabilistic features that are used to make predictions for novel ceRNAs of PTEN. Using a series of experiments in human prostate cancer cell lines, we validate the highest ranking prediction (TNRC6B) as a ceRNA of PTEN. The approach developed can be applied to map ceRNA networks of critical cellular regulators and to develop novel insights into crosstalk between different pathways involved in cancer.

Prediction of bacterial small RNAs in the RsmA (CsrA) and ToxT pathways: a machine learning approach

BackgroundSmall RNAs (sRNAs) constitute an important class of post-transcriptional regulators that control critical cellular processes in bacteria. Recent research using high-throughput transcriptomic approaches has led to a dramatic increase in the discovery of bacterial sRNAs. However, it is generally believed that the currently identified sRNAs constitute a limited subset of the bacterial sRNA repertoire. In several cases, sRNAs belonging to a specific class are already known and the challenge is to identify additional sRNAs belonging to the same class. In such cases, machine-learning approaches can be used to predict novel sRNAs in a given class.MethodsIn this work, we develop novel bioinformatics approaches that integrate sequence and structure-based features to train machine-learning models for the discovery of bacterial sRNAs. We show that features derived from recurrent structural motifs in the ensemble of low energy secondary structures can distinguish the RNA classes with high accuracy.ResultsWe apply this approach to predict new members in two broad classes of bacterial small RNAs: 1) sRNAs that bind to the RNA-binding protein RsmA/CsrA in diverse bacterial species and 2) sRNAs regulated by the master regulator of virulence, ToxT, in Vibrio cholerae.ConclusionThe involvement of sRNAs in bacterial adaptation to changing environments is an increasingly recurring theme in current research in microbiology. It is likely that future research, combining experimental and computational approaches, will discover many more examples of sRNAs as components of critical regulatory pathways in bacteria. We have developed a novel approach for prediction of small RNA regulators in important bacterial pathways. This approach can be applied to specific classes of sRNAs for which several members have been identified and the challenge is to identify additional sRNAs.

Robust clinical outcome prediction based on Bayesian analysis of transcriptional profiles and prior causal networks

Motivation: Understanding and predicting an individual’s response in a clinical trial is the key to better treatments and cost-effective medicine. Over the coming years, more and more large-scale omics datasets will become available to characterize patients with complex and heterogeneous diseases at a molecular level. Unfortunately, genetic, phenotypical and environmental variation is much higher in a human trial population than currently modeled or measured in most animal studies. In our experience, this high variability can lead to failure of trained predictors in independent studies and undermines the credibility and utility of promising high-dimensional datasets. Methods: We propose a method that utilizes patient-level genome-wide expression data in conjunction with causal networks based on prior knowledge. Our approach determines a differential expression profile for each patient and uses a Bayesian approach to infer corresponding upstream regulators. These regulators and their corresponding posterior probabilities of activity are used in a regularized regression framework to predict response. Results: We validated our approach using two clinically relevant phenotypes, namely acute rejection in kidney transplantation and response to Infliximab in ulcerative colitis. To demonstrate pitfalls in translating trained predictors across independent trials, we analyze performance characteristics of our approach as well as alternative feature sets in the regression on two independent datasets for each phenotype. We show that the proposed approach is able to successfully incorporate causal prior knowledge to give robust performance estimates. Contact: daniel.ziemek@pfizer.com Supplementary information: Supplementary data are available at Bioinformatics online.

Robust phenotype prediction from gene expression data using differential shrinkage of co-regulated genes

Discovery of robust diagnostic or prognostic biomarkers is a key to optimizing therapeutic benefit for select patient cohorts - an idea commonly referred to as precision medicine. Most discovery studies to derive such markers from high-dimensional transcriptomics datasets are weakly powered with sample sizes in the tens of patients. Therefore, highly regularized statistical approaches are essential to making generalizable predictions. At the same time, prior knowledge-driven approaches have been successfully applied to the manual interpretation of high-dimensional transcriptomics datasets. In this work, we assess the impact of combining two orthogonal approaches for the discovery of biomarker signatures, namely (1) well-known lasso-based regression approaches and its more recent derivative, the group lasso, and (2) the discovery of significant upstream regulators in literature-derived biological networks. Our method integrates both approaches in a weighted group-lasso model and differentially weights gene sets based on inferred active regulatory mechanism. Using nested cross-validation as well as independent clinical datasets, we demonstrate that our approach leads to increased accuracy and generalizable results. We implement our approach in a computationally efficient, user-friendly R package called creNET. The package can be downloaded at https://github.com/kouroshz/creNethttps://github.com/kouroshz/creNet and is accompanied by a parsed version of the STRING DB data base.