Guido Sanguinetti

Missing data in kernel PCA

Kernel Principal Component Analysis (KPCA) is a widely used technique for visualisation and feature extraction. Despite its success and flexibility, the lack of a probabilistic interpretation means that some problems, such as handling missing or corrupted data, are very hard to deal with. In this paper we exploit the probabilistic interpretation of linear PCA together with recent results on latent variable models in Gaussian Processes in order to introduce an objective function for KPCA. This in turn allows a principled approach to the missing data problem. Furthermore, this new approach can be extended to reconstruct corrupted test data using fixed kernel feature extractors. The experimental results show strong improvements over widely used heuristics.

Carbon Monoxide-releasing Antibacterial Molecules Target Respiration and Global Transcriptional Regulators

Carbon monoxide, a classical respiratory inhibitor, also exerts vasodilatory, anti-inflammatory, and antiapoptotic effects. CO-releasing molecules have therapeutic value, increasing phagocytosis and reducing sepsis-induced lethality. Here we identify for the first time the bacterial targets of Ru(CO)3Cl(glycinate) (CORM-3), a ruthenium-based carbonyl that liberates CO rapidly under physiological conditions. Contrary to the expectation that CO would be preferentially inhibitory at low oxygen tensions or anaerobically, Escherichia coli cultures were also sensitive to CORM-3 at concentrations equimolar with oxygen. CORM-3, assayed as ruthenium, was taken up by bacteria and rapidly delivered CO intracellularly to terminal oxidases. Microarray analysis of CORM-3-treated cells revealed extensively modified gene expression, notably down-regulation of genes encoding key aerobic respiratory complexes. Genes involved in metal metabolism, homeostasis, or transport were also differentially expressed, and free intracellular zinc levels were elevated. Probabilistic modeling of transcriptomic data identified the global transcription regulators ArcA, CRP, Fis, FNR, Fur, BaeR, CpxR, and IHF as targets and potential CO sensors. Our discovery that CORM-3 is an effective inhibitor and global regulator of gene expression, especially under aerobic conditions, has important implications for administration of CO-releasing agents in sepsis and inflammation.

Transition of Escherichia coli from Aerobic to Micro-aerobic Conditions Involves Fast and Slow Reacting Regulatory Components

Understanding life at a systems level is a major aim of biology. The bacterium Escherichia coli offers one of the best opportunities to achieve this goal. It is a metabolically versatile bacterium able to respond to changes in oxygen availability. This ability is a crucial component of its lifestyle, allowing it to thrive in aerobic external environments and under the oxygen-starved conditions of a host gut. The controlled growth conditions of chemostat culture were combined with transcript profiling to investigate transcriptome dynamics during the transition from aerobic to micro-aerobic conditions. In addition to predictable changes in transcripts encoding proteins of central metabolism, the abundances of transcripts involved in homeostasis of redox-reactive metals (Cu and Fe), and cell envelope stress were significantly altered. To gain further insight into the responses of the regulatory networks, the activities of key transcription factors during the transition to micro-aerobic conditions were inferred using a probabilistic modeling approach, which revealed that the response of the direct oxygen sensor FNR was rapid and overshot, whereas the indirect oxygen sensor ArcA reacted more slowly. Similarly, the cell envelope stress sensors RpoE and CpxR reacted rapidly and more slowly, respectively. Thus, it is suggested that combining rapid and slow reacting components in regulatory networks might be a feature of systems in which a signal is perceived by two or more functionally related transcription factors controlling overlapping regulons.

Probabilistic inference of transcription factor concentrations and gene-specific regulatory activities

MOTIVATION Quantitative estimation of the regulatory relationship between transcription factors and genes is a fundamental stepping stone when trying to develop models of cellular processes. Recent experimental high-throughput techniques, such as Chromatin Immunoprecipitation (ChIP) provide important information about the architecture of the regulatory networks in the cell. However, it is very difficult to measure the concentration levels of transcription factor proteins and determine their regulatory effect on gene transcription. It is therefore an important computational challenge to infer these quantities using gene expression data and network architecture data. RESULTS We develop a probabilistic state space model that allows genome-wide inference of both transcription factor protein concentrations and their effect on the transcription rates of each target gene from microarray data. We use variational inference techniques to learn the model parameters and perform posterior inference of protein concentrations and regulatory strengths. The probabilistic nature of the model also means that we can associate credibility intervals to our estimates, as well as providing a tool to detect which binding events lead to significant regulation. We demonstrate our model on artificial data and on two yeast datasets in which the network structure has previously been obtained using ChIP data. Predictions from our model are consistent with the underlying biology and offer novel quantitative insights into the regulatory structure of the yeast cell. AVAILABILITY MATLAB code is available from http://umber.sbs.man.ac.uk/resources/puma

A new probabilistic generative model of parameter inference in biochemical networks

We present a new method for estimating rate coefficients and level of noise in models of biochemical networks from noisy observations of concentration levels at discrete time points. Its probabilistic formulation, based on maximum likelihood estimation, is key to a principled handling of the noise inherent in biological data, and it allows for a number of further extensions, such as a fully Bayesian treatment of the parameter inference and automated model selection strategies based on the comparison between marginal likelihoods of different models. We developed KInfer (Knowlegde Inference), a tool implementing our inference model. KInfer is downloadable for free at http://www.cosbi.eu.

Point process modelling of the Afghan War Diary

Modern conflicts are characterized by an ever increasing use of information and sensing technology, resulting in vast amounts of high resolution data. Modelling and prediction of conflict, however, remain challenging tasks due to the heterogeneous and dynamic nature of the data typically available. Here we propose the use of dynamic spatiotemporal modelling tools for the identification of complex underlying processes in conflict, such as diffusion, relocation, heterogeneous escalation, and volatility. Using ideas from statistics, signal processing, and ecology, we provide a predictive framework able to assimilate data and give confidence estimates on the predictions. We demonstrate our methods on the WikiLeaks Afghan War Diary. Our results show that the approach allows deeper insights into conflict dynamics and allows a strikingly statistically accurate forward prediction of armed opposition group activity in 2010, based solely on data from previous years.

puma: a Bioconductor package for propagating uncertainty in microarray analysis

BackgroundMost analyses of microarray data are based on point estimates of expression levels and ignore the uncertainty of such estimates. By determining uncertainties from Affymetrix GeneChip data and propagating these uncertainties to downstream analyses it has been shown that we can improve results of differential expression detection, principal component analysis and clustering. Previously, implementations of these uncertainty propagation methods have only been available as separate packages, written in different languages. Previous implementations have also suffered from being very costly to compute, and in the case of differential expression detection, have been limited in the experimental designs to which they can be applied.Resultspuma is a Bioconductor package incorporating a suite of analysis methods for use on Affymetrix GeneChip data. puma extends the differential expression detection methods of previous work from the 2-class case to the multi-factorial case. puma can be used to automatically create design and contrast matrices for typical experimental designs, which can be used both within the package itself but also in other Bioconductor packages. The implementation of differential expression detection methods has been parallelised leading to significant decreases in processing time on a range of computer architectures. puma incorporates the first R implementation of an uncertainty propagation version of principal component analysis, and an implementation of a clustering method based on uncertainty propagation. All of these techniques are brought together in a single, easy-to-use package with clear, task-based documentation.ConclusionFor the first time, the puma package makes a suite of uncertainty propagation methods available to a general audience. These methods can be used to improve results from more traditional analyses of microarray data. puma also offers improvements in terms of scope and speed of execution over previously available methods. puma is recommended for anyone working with the Affymetrix GeneChip platform for gene expression analysis and can also be applied more generally.

Transcript Profiling and Inference of Escherichia coli K-12 ArcA Activity across the Range of Physiologically Relevant Oxygen Concentrations

Oxygen availability is the major determinant of the metabolic modes adopted by Escherichia coli. Although much is known about E. coli gene expression and metabolism under fully aerobic and anaerobic conditions, the intermediate oxygen tensions that are encountered in natural niches are understudied. Here, for the first time, the transcript profiles of E. coli K-12 across the physiologically significant range of oxygen availabilities are described. These suggested a progressive switch to aerobic respiratory metabolism and a remodeling of the cell envelope as oxygen availability increased. The transcriptional responses were consistent with changes in the abundance of cytochrome bd and bo′ and the outer membrane protein OmpW. The observed transcript and protein profiles result from changes in the activities of regulators that respond to oxygen itself or to metabolic and environmental signals that are sensitive to oxygen availability (aerobiosis). A probabilistic model (TFInfer) was used to predict the activity of the indirect oxygen-sensing two-component system ArcBA across the aerobiosis range. The model implied that the activity of the regulator ArcA correlated with aerobiosis but not with the redox state of the ubiquinone pool, challenging the idea that ArcA activity is inhibited by oxidized ubiquinone. The amount of phosphorylated ArcA correlated with the predicted ArcA activities and with aerobiosis, suggesting that fermentation product-mediated inhibition of ArcB phosphatase activity is the dominant mechanism for regulating ArcA activity under the conditions used here.

Automatic Determination of the Number of Clusters Using Spectral Algorithms

We introduce a novel spectral clustering algorithm that allows us to automatically determine the number of clusters in a dataset. The algorithm is based on a theoretical analysis of the spectral properties of block diagonal affinity matrices; in contrast to established methods, we do not normalise the rows of the matrix of eigenvectors, and argue that the non-normalised data contains key information that allows the automatic determination of the number of clusters present. We present several examples of datasets successfully clustered by our algorithm, both artificial and real, obtaining good results even without employing refined feature extraction techniques

Peroxynitrite Toxicity in Escherichia coli K12 Elicits Expression of Oxidative Stress Responses and Protein Nitration and Nitrosylation

Peroxynitrite is formed in macrophages by the diffusion-limited reaction of superoxide and nitric oxide. This highly reactive species is thought to contribute to bacterial killing by interaction with diverse targets and nitration of protein tyrosines. This work presents for the first time a comprehensive analysis of transcriptional responses to peroxynitrite under tightly controlled chemostat growth conditions. Up-regulation of the cysteine biosynthesis pathway and an increase in S-nitrosothiol levels suggest S-nitrosylation to be a consequence of peroxynitrite exposure. Genes involved in the assembly/repair of iron-sulfur clusters also show enhanced transcription. Unexpectedly, arginine biosynthesis gene transcription levels were also elevated after treatment with peroxynitrite. Analysis of the negative regulator for these genes, ArgR, showed that post-translational nitration of tyrosine residues within this protein is responsible for its degradation in vitro. Further up-regulation was seen in oxidative stress response genes, including katG and ahpCF. However, genes known to be up-regulated by nitric oxide and nitrosating agents (e.g. hmp and norVW) were unaffected. Probabilistic modeling of the transcriptomic data identified five altered transcription factors in response to peroxynitrite exposure, including OxyR and ArgR. Hydrogen peroxide can be present as a contaminant in commercially available peroxynitrite preparations. Transcriptomic analysis of cells treated with hydrogen peroxide alone also revealed up-regulation of oxidative stress response genes but not of many other genes that are up-regulated by peroxynitrite. Thus, the cellular responses to peroxynitrite and hydrogen peroxide are distinct.