Heather Turner

Improved biclustering of microarray data demonstrated through systematic performance tests

A new algorithm is presented for fitting the plaid model, a biclustering method developed for clustering gene expression data. The approach is based on speedy individual differences clustering and uses binary least squares to update the cluster membership parameters, making use of the binary constraints on these parameters and simplifying the other parameter updates. The performance of both algorithms is tested on simulated data sets designed to imitate (normalised) gene expression data, covering a range of biclustering configurations. Empirical distributions for the components of these data sets, including non-systematic error, are derived from a real set of microarray data. A set of two-way quality measures is proposed, based on one-way measures commonly used in information retrieval, to evaluate the quality of a retrieved bicluster with respect to a target bicluster in terms of both genes and samples. By defining a one-to-one correspondence between target biclusters and retrieved biclusters, the performance of each algorithm can be assessed. The results show that, using appropriately selected starting criteria, the proposed algorithm out-performs the original plaid model algorithm across a range of data sets. Furthermore, through the rigorous assessment of the plaid model a benchmark for future evaluation of biclustering methods is established.

Biclustering Models for Structured Microarray Data

Microarrays have become a standard tool for investigating gene function and more complex microarray experiments are increasingly being conducted. For example, an experiment may involve samples from several groups or may investigate changes in gene expression over time for several subjects, leading to large three-way data sets. In response to this increase in data complexity, we propose some extensions to the plaid model, a biclustering method developed for the analysis of gene expression data. This model-based method lends itself to the incorporation of any additional structure such as external grouping or repeated measures. We describe how the extended models may be fitted and illustrate their use on real data.

Deciphering c-MYC-regulated genes in two distinct tissues

BackgroundThe transcription factor MYC is a critical regulator of diverse cellular processes, including both replication and apoptosis. Differences in MYC-regulated gene expression responsible for such opposing outcomes in vivo remain obscure. To address this we have examined time-dependent changes in global gene expression in two transgenic mouse models in which MYC activation, in either skin suprabasal keratinocytes or pancreatic islet β-cells, promotes tissue expansion or involution, respectively.ResultsConsistent with observed phenotypes, expression of cell cycle genes is increased in both models (albeit enriched in β-cells), as are those involved in cell growth and metabolism, while expression of genes involved in cell differentiation is down-regulated. However, in β-cells, which unlike suprabasal keratinocytes undergo prominent apoptosis from 24 hours, there is up-regulation of genes associated with DNA-damage response and intrinsic apoptotic pathways, including Atr, Arf, Bax and Cycs. In striking contrast, this is not the case for suprabasal keratinocytes, where pro-apoptotic genes such as Noxa are down-regulated and key anti-apoptotic pathways (such as Igf1-Akt) and those promoting angiogenesis are up-regulated. Moreover, dramatic up-regulation of steroid hormone-regulated Kallikrein serine protease family members in suprabasal keratinocytes alone could further enhance local Igf1 actions, such as through proteolysis of Igf1 binding proteins.ConclusionsActivation of MYC causes cell growth, loss of differentiation and cell cycle entry in both β-cells and suprabasal keratinocytes in vivo. Apoptosis, which is confined to β-cells, may involve a combination of a DNA-damage response and downstream activation of pro-apoptotic signalling pathways, including Cdc2a and p19Arf/p53, and downstream targets. Conversely, avoidance of apoptosis in suprabasal keratinocytes may result primarily from the activation of key anti-apoptotic signalling pathways, particularly Igf1-Akt, and induction of an angiogenic response, though intrinsic resistance to induction of p19Arf by MYC in suprabasal keratinocytes may contribute.

BIGL: Biochemically Intuitive Generalized Loewe null model for prediction of the expected combined effect compatible with partial agonism and antagonism

Clinical efficacy regularly requires the combination of drugs. For an early estimation of the clinical value of (potentially many) combinations of pharmacologic compounds during discovery, the observed combination effect is typically compared to that expected under a null model. Mechanistic accuracy of that null model is not aspired to; to the contrary, combinations that deviate favorably from the model (and thereby disprove its accuracy) are prioritized. Arguably the most popular null model is the Loewe Additivity model, which conceptually maps any assay under study to a (virtual) single-step enzymatic reaction. It is easy-to-interpret and requires no other information than the concentration-response curves of the individual compounds. However, the original Loewe model cannot accommodate concentration-response curves with different maximal responses and, by consequence, combinations of an agonist with a partial or inverse agonist. We propose an extension, named Biochemically Intuitive Generalized Loewe (BIGL), that can address different maximal responses, while preserving the biochemical underpinning and interpretability of the original Loewe model. In addition, we formulate statistical tests for detecting synergy and antagonism, which allow for detecting statistically significant greater/lesser observed combined effects than expected from the null model. Finally, we demonstrate the novel method through application to several publicly available datasets.

Disability weights for infectious diseases in four European countries : comparison between countries and across respondent characteristics

Abstract Background In 2015, new disability weights (DWs) for infectious diseases were constructed based on data from four European countries. In this paper, we evaluated if country, age, sex, disease experience status, income and educational levels have an impact on these DWs. Methods We analyzed paired comparison responses of the European DW study by participants’ characteristics with separate probit regression models. To evaluate the effect of participants’ characteristics, we performed correlation analyses between countries and within country by respondent characteristics and constructed seven probit regression models, including a null model and six models containing participants’ characteristics. We compared these seven models using Akaike Information Criterion (AIC). Results According to AIC, the probit model including country as covariate was the best model. We found a lower correlation of the probit coefficients between countries and income levels (range rs: 0.97–0.99, P < 0.01) than between age groups (range rs: 0.98–0.99, P < 0.01), educational level (range rs: 0.98–0.99, P < 0.01), sex (rs = 0.99, P < 0.01) and disease status (rs = 0.99, P < 0.01). Within country the lowest correlations of the probit coefficients were between low and high income level (range rs = 0.89–0.94, P < 0.01). Conclusions We observed variations in health valuation across countries and within country between income levels. These observations should be further explored in a systematic way, also in non-European countries. We recommend future researches studying the effect of other characteristics of respondents on health assessment.