Search

This paper reviews recent computational approaches to the understanding of gene regulation in eukaryotes. Cis-regulation of gene expression by the binding of transcription factors is a critical component of cellular physiology. In eukaryotes, a number of transcription factors often work together in a combinatorial fashion to enable cells to respond to a wide spectrum of environmental and developmental signals. Integration of genome sequences and/or Chromatin Immunoprecipitation on chip data with gene-expression data has facilitated in silico discovery of how the combinatorics and positioning of transcription factors binding sites underlie gene activation in a variety of cellular processes. The process of gene regulation is extremely complex and intriguing, therefore all possible points of view and related links should be carefully considered. Here we attempt to collect an inventory, not claiming it to be comprehensive and complete, of related computational biological topics covering gene regulation, which may en-lighten the process, and briefly review what is currently occurring in these areas. We will consider the following computational areas: o gene regulatory network construction; o evolution of regulatory DNA; o studies of its structural and statistical informational properties; o and finally, regulatory RNA.

Parkinson's disease (PD) is a common, progressive, incurable disabling condition. The cause is unknown but over the past few years tremendous progress in our understanding of the genetic bases of this condition has been made. To date, this has almost exclusively come from the study of relatively rare Mendelian forms of the disease and there are no currently, widely accepted common variants known to increase susceptibility.The role that the ‘Mendelian’ genes play in common sporadic forms of PD is unknown. Moreover, most studies in PD can really be described as candidate polymorphism studies rather than true and complete assessments of the genes themselves. We provide a model of how one might tackle some of these issues using Parkinson's disease as an illustration. One of the emerging hypotheses of gene environment interaction in Parkinson's disease is based on drug metabolizing (or xenobiotic) enzymes and their interaction with putative environmental toxins. This motivated us to describe a tagging approach for an extensive but not exhaustive list of 55 drug metabolizing enzyme genes. We use these data to illustrate the power, and some of the limitations of a haplotype tagging approach. We show that haplotype tagging is extremely efficient and works well with only a modest increase in effort through different populations. The tagging approach works much less well if the minor allele frequency is below 5%. However, it will now be possible using these tags to evaluate these genes comprehensively in PD and other neurodegenerative conditions.

Many studies have shown long-term benefits of HLA matching in kidney transplantation. By examining rates of graft loss within consecutive posttransplant intervals, using data from the UK Transplant Service, we show that the long-term benefits of HLA matching are due to reduction of the graft failure rate within five months of transplantation. After 5-months, HLA-A,B,DR matching appears to have little impact on graft loss. We suggest that graft losses after 5 months may be attributable to non-HLA targets.

Markov chain Monte Carlo (MCMC) techniques, such as the Gibbs sampler, are increasingly being used for Bayesian inference. We propose a new MCMC method: Adaptive direction sampling (ADS) which, unlike the Gibbs sampler, involves sampling in directions which adapt to the target density. We present non‐technically the essence of ADS, but with sufficient detail to allow the practitioner to apply the method. We demonstrate irreducibility of the snooker algorithm, a special case of ADS. We compare the performance of special cases of ADS, including the snooker algorithm and the Gibbs sampler, in a simple test example.

Gibbs sampling has enormous potential for analysing complex data sets. However, routine use of Gibbs sampling has been hampered by the lack of general purpose software for its implementation. Until now all applications have involved writing one‐off computer code in low or intermediate level languages such as C or Fortran. We describe some general purpose software that we are currently developing for implementing Gibbs sampling: BUGS (Bayesian inference using Gibbs sampling). The BUGS system comprises three components: first, a natural language for specifying complex models; second, an ‘expert system’ for deciding appropriate methods for obtaining samples required by the Gibbs sampler; third, a sampling module containing numerical routines to perform the sampling. S objects are used for data input and output. BUGS is written in Modula‐2 and runs under both DOS and UNIX.

Large data sets often contain many details of interest and importance which can only be elicited through exhaustive statistical modelling. This may not be practical because of computational expense involved in repeatedly processing the full data; the need to run analyses in off‐peak periods to reduce overheads; or limitations on the amount of data that can be accessed simultaneously. A rapid two‐stage modelling technique for exploring large data sets (Gilks, 1986) is described which, at the first stage, performs separate Maximum Likelihood (ML) estimations in one or more ‘segments’ of data and, at the second‐stage, combines these estimates to explore a wide range of models of the full data, under the direction of the investigator. The second‐stage estimates closely resemble ML estimates and in practice have been obtained in 1/300th of the computer time needed for an exact ML estimation. The method is illustrated with a structured, four‐stage analysis of a large tuberculosis vaccines clinical trial, and practical issues that have arisen during this and other applications of the method are discussed.

We describe Bayesian hierarchical-spatial models for disease mapping with imprecisely observed ecological covariates. We posit smoothing priors for both the disease submodel and the covariate submodel. We apply the models to an analysis of insulin Dependent Diabetes Mellitus incidence in Sardinia, with malaria prevalence as a covariate. © 1997 by John Wiley & Sons, Ltd.

Risk factors used in epidemiology are often measured with error which can seriously affect the assessment of the relation between risk factors and disease outcome. In this paper, a Bayesian perspective on measurement error problems in epidemiology is taken and it is shown how the information available in this setting can be structured in terms of conditional independence models. The modeling of common designs used in the presence of measurement error (validation group, repeated measures, ancillary data) is described. The authors indicate how Bayesian estimation can be carried out in these settings using Gibbs sampling, a sampling technique which is being increasingly referred to in statistical and biomedical applications. The method is illustrated by analyzing a design with two measuring instruments and no validation group.

The number of acquired immune deficiency syndrome (AIDS) cases reported in England and Wales in the recent past seriously underestimates the number of recent AIDS diagnoses, because of substantial reporting delays. We examine two methods for estimating AIDS diagnoses, taking account of reporting delays. The first is a method based on conditional likelihood making minimal parametric assumptions, with confidence intervals assessed through bootstrapping. The second is a fully parametric Bayesian approach estimated by Gibbs sampling. We apply these methods to AIDS reports in England and Wales from July 1983 to December 1992, and we compare the results.

Sixty patients who presented for day‐case dilatation and curettage were allocated randomly to receive either thiopentone or propofol for induction and maintenance of anaesthesia. One anaesthetist administered all the anaesthetics whilst all assessments were made by one other. The results indicate that early recovery of memory function, critical flicker fusion frequency and subjective feelings of tiredness, drowsiness and alertness were superior in the propofol group. There was a significant difference in subjective feelings of tiredness and drowsiness recorded by the two study groups at 24 hours. Memory function assessed by Wechsler logical memory function passages at 24 hours was impaired in the propofol group in comparison to a group of ’reference’ subjects.