Martino Barenco

Ranked prediction of p53 targets using hidden variable dynamic modeling

Full exploitation of microarray data requires hidden information that cannot be extracted using current analysis methodologies. We present a new approach, hidden variable dynamic modeling (HVDM), which derives the hidden profile of a transcription factor from time series microarray data, and generates a ranked list of predicted targets. We applied HVDM to the p53 network, validating predictions experimentally using small interfering RNA. HVDM can be applied in many systems biology contexts to predict regulation of gene activity quantitatively.

Fitting ordinary differential equations to short time course data

Ordinary differential equations (ODEs) are widely used to model many systems in physics, chemistry, engineering and biology. Often one wants to compare such equations with observed time course data, and use this to estimate parameters. Surprisingly, practical algorithms for doing this are relatively poorly developed, particularly in comparison with the sophistication of numerical methods for solving both initial and boundary value problems for differential equations, and for locating and analysing bifurcations. A lack of good numerical fitting methods is particularly problematic in the context of systems biology where only a handful of time points may be available. In this paper, we present a survey of existing algorithms and describe the main approaches. We also introduce and evaluate a new efficient technique for estimating ODEs linear in parameters particularly suited to situations where noise levels are high and the number of data points is low. It employs a spline-based collocation scheme and alternates linear least squares minimization steps with repeated estimates of the noise-free values of the variables. This is reminiscent of expectation–maximization methods widely used for problems with nuisance parameters or missing data.

Dissection of a complex transcriptional response using genome-wide transcriptional modelling

Modern genomics technologies generate huge data sets creating a demand for systems level, experimentally verified, analysis techniques. We examined the transcriptional response to DNA damage in a human T cell line (MOLT4) using microarrays. By measuring both mRNA accumulation and degradation over a short time course, we were able to construct a mechanistic model of the transcriptional response. The model predicted three dominant transcriptional activity profiles—an early response controlled by NFκB and c‐Jun, a delayed response controlled by p53, and a late response related to cell cycle re‐entry. The method also identified, with defined confidence limits, the transcriptional targets associated with each activity. Experimental inhibition of NFκB, c‐Jun and p53 confirmed that target predictions were accurate. Model predictions directly explained 70% of the 200 most significantly upregulated genes in the DNA‐damage response. Genome‐wide transcriptional modelling (GWTM) requires no prior knowledge of either transcription factors or their targets. GWTM is an economical and effective method for identifying the main transcriptional activators in a complex response and confidently predicting their targets.

rHVDM – a fast and user-friendly R package to predict transcription factor targets from microarray time series data

Address: 1Institute of Child Health, University College London, WC1N 1EH, UK, 2Bloomsbury Centre for Bioinformatics, University College London, WC1E 6BT, UK, 3Institute of Cancer Research, Sutton, SM2 5NG, UK, 4Department of Mathematics, Imperial College, London, SW7 2BZ, UK, 5Patterson Institute for Cancer Research, University of Manchester, M20 4BX, UK and 6CoMPLEX, University College London, NW1 2HE, UK

Correction of scaling mismatches in oligonucleotide microarray data

BackgroundGene expression microarray data is notoriously subject to high signal variability. Moreover, unavoidable variation in the concentration of transcripts applied to microarrays may result in poor scaling of the summarized data which can hamper analytical interpretations. This is especially relevant in a systems biology context, where systematic biases in the signals of particular genes can have severe effects on subsequent analyses. Conventionally it would be necessary to replace the mismatched arrays, but individual time points cannot be rerun and inserted because of experimental variability. It would therefore be necessary to repeat the whole time series experiment, which is both impractical and expensive.ResultsWe explain how scaling mismatches occur in data summarized by the popular MAS5 (GCOS; Affymetrix) algorithm, and propose a simple recursive algorithm to correct them. Its principle is to identify a set of constant genes and to use this set to rescale the microarray signals. We study the properties of the algorithm using artificially generated data and apply it to experimental data. We show that the set of constant genes it generates can be used to rescale data from other experiments, provided that the underlying system is similar to the original. We also demonstrate, using a simple example, that the method can successfully correct existing imbalancesin the data.ConclusionThe set of constant genes obtained for a given experiment can be applied to other experiments, provided the systems studied are sufficiently similar. This type of rescaling is especially relevant in systems biology applications using microarray data.

A genomic atlas of human adrenal and gonad development

Background: In humans, the adrenal glands and gonads undergo distinct biological events between 6-10 weeks post conception (wpc), such as testis determination, the onset of steroidogenesis and primordial germ cell development. However, relatively little is currently known about the genetic mechanisms underlying these processes. We therefore aimed to generate a detailed genomic atlas of adrenal and gonad development across these critical stages of human embryonic and fetal development. Methods: RNA was extracted from 53 tissue samples between 6-10 wpc (adrenal, testis, ovary and control). Affymetrix array analysis was performed and differential gene expression was analysed using Bioconductor. A mathematical model was constructed to investigate time-series changes across the dataset. Pathway analysis was performed using ClueGo and cellular localisation of novel factors confirmed using immunohistochemistry. Results: Using this approach, we have identified novel components of adrenal development (e.g. ASB4, NPR3) and confirmed the role of SRY as the main human testis-determining gene. By mathematical modelling time-series data we have found new genes up-regulated with SOX9 in the testis (e.g. CITED1), which may represent components of the testis development pathway. We have shown that testicular steroidogenesis has a distinct onset at around 8 wpc and identified potential novel components in adrenal and testicular steroidogenesis (e.g. MGARP, FOXO4, MAP3K15, GRAMD1B, RMND2), as well as testis biomarkers (e.g. SCUBE1). We have also shown that the developing human ovary expresses distinct subsets of genes (e.g. OR10G9, OR4D5), but enrichment for established biological pathways is limited. Conclusion: This genomic atlas is revealing important novel aspects of human development and new candidate genes for adrenal and reproductive disorders.

A genome wide transcriptional model of the complex response to pre-TCR signalling during thymocyte differentiation.

Developing thymocytes require pre-TCR signalling to differentiate from CD4−CD8− double negative to CD4+CD8+ double positive cell. Here we followed the transcriptional response to pre-TCR signalling in a synchronised population of differentiating double negative thymocytes. This time series analysis revealed a complex transcriptional response, in which thousands of genes were up and down-regulated before changes in cell surface phenotype were detected. Genome-wide measurement of RNA degradation of individual genes showed great heterogeneity in the rate of degradation between different genes. We therefore used time course expression and degradation data and a genome wide transcriptional modelling (GWTM) strategy to model the transcriptional response of genes up-regulated on pre-TCR signal transduction. This analysis revealed five major temporally distinct transcriptional activities that up regulate transcription through time, whereas down-regulation of expression occurred in three waves. Our model thus placed known regulators in a temporal perspective, and in addition identified novel candidate regulators of thymocyte differentiation.

Revealing Individual Signatures of Human T Cell CDR3 Sequence Repertoires with Kidera Factors

The recent development of High Throughput Sequencing technologies has enabled an individual’s TCR repertoire to be efficiently analysed at the nucleotide level. However, with unique clonotypes ranging in the tens of millions per individual, this approach gives a surfeit of information that is difficult to analyse and interpret in a biological context and gives little information about TCR structural diversity. Using publicly available TCR CDR3 sequence data, we analysed TCR repertoires by converting the encoded CDR3 amino acid sequences into Kidera Factors, a set of orthogonal physico-chemical properties that reflect protein structure. This approach enabled the TCR repertoire from different individuals to be distinguished and demonstrated the close similarity of the repertoire in different samples from the same individual.

rHVDM: an R package to predict the activity and targets of a transcription factor

SUMMARY Highly parallel genomic platforms like microarrays often present researchers with long lists of differentially expressed genes but contain little or no information on how these genes are regulated. rHVDM is a novel R package which uses gene expression time course data to predict the activity and targets of a transcription factor. In the first step, rHVDM uses a small number of known targets to derive the activity profile of a given transcription factor. Then, in a subsequent step, this activity profile is used to predict other putative targets of that transcription factor. A dynamic and mechanistic model of gene expression is at the heart of the technique. Measurement error is taken into account during the process, which allows an objective assessment of the robustness of fit and, therefore, the quality of the predictions. The package relies on efficient algorithms and vectorization to accomplish potentially time consuming tasks including optimization and differential equation integration. We demonstrate the efficiency and accuracy of rHVDM by examining the activity of the tumour-suppressing transcription factor, p53. AVAILABILITY The version of the package presented here (1.8.1) is freely available from the Bioconductor Web site (http://bioconductor.org/packages/2.3/bioc/html/rHVDM.html).

IFITM proteins drive type 2 T helper cell differentiation and exacerbate allergic airway inflammation

The interferon‐inducible transmembrane (Ifitm/Fragilis) genes encode homologous proteins that are induced by IFNs. Here, we show that IFITM proteins regulate murine CD4+ Th cell differentiation. Ifitm2 and Ifitm3 are expressed in wild‐type (WT) CD4+ T cells. On activation, Ifitm3 was downregulated and Ifitm2 was upregulated. Resting Ifitm‐family‐deficient CD4+ T cells had higher expression of Th1‐associated genes than WT and purified naive Ifitm‐family‐deficient CD4+ T cells differentiated more efficiently to Th1, whereas Th2 differentiation was inhibited.