Anthony Rossini

Bioconductor: open software development for computational biology and bioinformatics

The Bioconductor project is an initiative for the collaborative creation of extensible software for computational biology and bioinformatics. The goals of the project include: fostering collaborative development and widespread use of innovative software, reducing barriers to entry into interdisciplinary scientific research, and promoting the achievement of remote reproducibility of research results. We describe details of our aims and methods, identify current challenges, compare Bioconductor to other open bioinformatics projects, and provide working examples.

Overcoming barriers to HIV testing: preferences for new strategies among clients of a needle exchange, a sexually transmitted disease clinic, and sex venues for men who have sex with men:

Objective: To determine strategies to overcome barriers to HIV testing among persons at risk. Methods: We developed a survey that elicited testing motivators, barriers, and preferences for new strategies among 460 participants at a needle exchange, three sex venues for men who have sex with men, and a sexually transmitted disease clinic. Results: Barriers to testing included factors influenced by individual concern (fear and discrimination); by programs, policies, and laws (named reporting and inability to afford treatment); and by counseling and testing strategies (dislike of counseling, anxiety waiting for results, and venipuncture). The largest proportions of participants preferred rapid testing strategies, including clinic‐based testing (27%) and home selftesting (20%); roughly equal proportions preferred oral fluid testing (18%), urine testing (17%), and standard blood testing (17%). One percent preferred home specimen collection. Participants who had never tested before were significantly more likely to prefer home self‐testing compared with other strategies. Blacks were significantly more likely to prefer urine testing. Conclusions: Strategies for improving acceptance of HIV counseling and testing include information about access to anonymous testing and early treatment. Expanding options for rapid testing, urine testing, and home self‐testing; providing alternatives to venipuncture; making pretest counseling optional; and allowing telephone results disclosure may encourage more persons to learn their HIV status.

Expression Profiling Identifies Smooth Muscle Cell Diversity Within Human Intima and Plaque Fibrous Cap: Loss of RGS5 Distinguishes the Cap

Background—The fibrous cap of the atherosclerotic lesion is believed to be critical to stability because disruption of the cap is the final event leading to plaque rupture. We have, therefore, used expression arrays to define the phenotype of the cap and other plaque components. Methods and Results—To identify unique expression programs able to distinguish the smooth muscle of the cap from other plaque smooth muscle cells, RNA profiles were determined in human carotid artery media, nonatherosclerotic adjacent intima, fibrous cap of advanced atherosclerotic plaques, and whole advanced plaque with cDNA arrays covering 21 000 or 26 000 Unigene clusters. The molecular signature of each tissue was dominated by a core gene-set with differential expression of <1% of clusters assayed. Conclusions—Both intima and cap expressed novel genes not previously associated with SMC pathology. If the cap is derived from a unique subpopulation, this pattern is the signature of that particular set of cells. The loss of RGS5 in the fibrous cap is of particular interest because of its role in vessel development and physiology.

Orca: A Visualization Toolkit for High-Dimensional Data

Abstract This article describes constructing interactive and dynamic linked data views using the Java programming language. The data views are designed for data that have a multivariate component. The approach to displaying data comes from earlier research on building statistical graphics based on data pipelines, in which different aspects of data processing and graphical rendering are organized conceptually into segments of a pipeline. The software design takes advantage of the object-oriented nature of the Java language to open up the data pipeline, allowing developers to have greater control over their visualization applications. Importantly, new types of data views coded to adhere to a few simple design requirements can easily be integrated with existing pipe sections. This allows access to sophisticated linking and dynamic interaction across all (new and existing) view types. Pipe segments can be accessed from data analysis packages such as Omegahat or R, providing a tight coupling of visual and numerical methods.

Emacs Speaks Statistics: A Multiplatform, Multipackage Development Environment for Statistical Analysis

Computer programming is an important component of statistical research and data analysis. It is a necessary skill for using sophisticated statistical packages and for writing custom scripts and software to perform data analysis using modern statistical methods. Emacs Speaks Statistics (ESS) provides an intelligent and consistent interface between the user and statistics software. ESS interfaces with SAS, S-Plus, R, and other statistics packages under the Unix, Microsoft Windows, and Apple Macintosh operating systems. ESS extends the Emacs text editor to streamline the use and creation of statistical software. ESS understands the syntax for numerous data analysis languages, provides consistent display and editing features across packages, and assists in the interactive or batch execution of statements by statistics packages. We describe in detail the features that ESS provides to increase efficiency.

Computable Statistical Research and Practice

The work of research and applied statisticians is driven by both manual and electronic computation. This computation, which supports our work, can be crudely described by 3 stages: scoping, to assess what can and needs to be done; analysis, where this is accomplished; and reporting, which communicates the results to others. Barriers to the reuse of computations can be found in the translational needs driving the transition between sub-activities; for example, scoping activities are seldom directly reusable during analysis, and there is a limited amount of direct reuse going from analysis to reporting. There is an additional high barrier for translating statistical theory and methodology to practical activities, with both sides (applied and theoretical statisticians) pointing the blame at the other for not using appropriate tools or addressing appropriate needs. In this sense, much statistical research is not really computable, but rather, translatable. This work describes some proposals for exploring novel information technology support to address the translational challenges during transition between stages of statistical practice. These are illustrated in the on-going design of CommonLisp Stat, a platform for investigating the interaction between statistical computing environments supporting research and/or practice.

On the Edge: Statistics & Computing: Bioterrorism Surveillance Using Computerized Clinical Data

The September II attacks transformed Americas perception of her relationship to the world, as well as her perception of her own vulnerability.The events in the months that followed had an even greater impact on the public health community. The report of the first case of inhalation anthrax, and the conclusion that the case was a result of intentional spread, demonstrated vividly the susceptibility of the population to bioterrorism. Though surveillance and outbreak investigation have long been central to public health, research and operational activities moved quickly to center stage for both public health and national security. With the increased focus on surveillance comes the need to incorporate novel sources of surveillance data. Extracting, integrating, and understanding the utility of surveillance data from heterogeneous sources remains a substantial challenge.

On the Edge: Statistics & Computing: “Agile” GIS: Building Application- Specific Spatial Analytic Software from Freely Available Software Tools

(:eographic information systems (( :ISs) provide a collection of software tools for managing, linking, and displaying spatially-referenced data but often offer limited options in the wav of statistical or spatial statistical analysis. Statistical software packages ollcr a wide range of analytical techniques, but lillie in the way of spatial data management. querying, linking, and analytical methods for spatial data analysis. While recent development of add-on modules or stand-alone packages offers some interconnection \wtwlTn the two worlds, emphasis seems to focus on adding the techniques of one set of software [0 the other. Both CIS and stat istical packages are large software environments providing tools for a wide spec! rum of tasks, of which any single application is likely to use onlv a subset. In contrast to a strategy of making big packages bigger. we usc three specific applications to illustrate an alternative approach built on the development of application-specific software built from freely available sets of both C;IS and statistical software components, The goal of developing such

Visualising high-dimensional data in time and space: ideas from the Orca project

Abstract Environmental data are frequently high-dimensional with measurements of multiple chemical constituents, plant or animal species, or meteorological variables. Environmental data are also frequently structured with interest in the patterns of variation over time and space. We describe some new data visualization methods from the Orca project that allow the analyst to reduce the dimension of the data without obscuring its basic structure and illustrate these on air pollution data.

On the Edge: Statistics & Computing: Something in the Air?

On May 30, 2002, air pollution regulators and researchers had a nasty surprise. The National Mortality and Morbidity Air Pollution Study (NMMAPS), a huge project jointly funded by industry and the Environmental Protection Agency (EPA) to examine health effects of air pollution in the largest 90 cities in the United States, revised its previouslypublished estimate of these effects (Samet et al. 2000). The revision decreased estimates by nearly one-half. For perhaps the first time ever, statistical computing problems made the front page of the New York Times. The NMMAPS researchers describe these problems in the August 1 issue of American Journal ofEpidemiology (Dominici et al. 2002), and a session at the 2002 Joint Statistical Meetings discussed the general impact of NMMAPS.