An article in the November issue of IEEE Computer Magazine (Sloot et al., From Molecule to Man: Decision Support in Individualized E-Health) turned me on to Virolab. This is a European Union-funded prototype "virtual laboratory" that the project page describes as follows:
...a virtual laboratory will be developed enabling easy access to distributed resources as well as sharing, processing and analysing virological, immunological, clinical and experimental data. As a prototype for this virtual laboratory for infectious diseases, the problem of HIV drug resistance will be used. The virtual laboratory will integrate the biomedical information from viruses (proteins and mutations), patients (e.g. viral load) and literature (drug resistance experiments) resulting in a rule-based distributed decision support system for drug ranking. In addition ViroLab will include advanced tools for (bio) statistical analysis, visualization, modelling and simulation, enabling prediction of the temporal virological and immunological response of viruses with complex mutation patterns for drug therapy.
The system seems to me to be very well thought through. It is targeted initially at a very important medical challenge, but this single focus is an artifact of its being a proof-of-concept project; the underlying technology is generically applicable to all manner of challenges in medical research and clinical care.
Virolab is an application of Web Services and grid technologies to the dual challenge of Type I and Type II translation. Type I translation takes knowledge from bench to bedside, from the laboratory to the hospital (in practice, to tertiary and quaternary care hospitals and academic health centers); Type II translates that knowledge from the domain of the specialist clinician to the general practice of medicine in the clinic and other community settings. The authors give a very concise statement of the driving forces behind the move toward medical applications of systems biology we are witnessing and participating in today:
An application pull has occurred in biomedicine with the move to in silico studies, which augment in vivo and in vitro studies by simulating more details of biomedical processes. Using these simulated processes helps medical doctors make decisions by exploring different scenarios. Preoperative simulation and visualization of vascular surgery and expert systems for drug ranking are two examples of such processes.
At the same time, a technology push is occurring in computing resources and data availability. In the field of high-performance computing, as computing advanced from sequential to parallel to distributed, killer applications moved from mathematics to physics, chemistry, biology, and now to medicine. In addition, advances in Internet technology and grid computing have made huge amounts of data available from sensors, experiments, and simulations. Still, significant computational, integration, collaboration, and interaction gaps exist between the observed application pull and the technology push. [Italics mine]
I found this last sentence very insightful. The computational and integrative challenges are well-known and widely publicized at this point, coinciding as they do with advances in systems engineering and commercial technology deployments. The latter two elements, collaboration and interaction, are harder to approach because they are rooted in the messy, vague, touchy-feely, not-necessarily-profitable domains of human-computer interaction (HCI) and computer-supported collaborative work (CSCW).
Virolab addresses these issues from a technological perspective, and this is very encouraging. Most of the health IT systems to which I am exposed do not provide a means for non-programmers to configure workflows. The article states the problem as follows, along with a hint of the solution:
ViroLab workflows enable dynamic workflow execution, lazy scheduling, and runtime recomposition. They also support two levels of abstraction needed to operate separately on abstract workflows (workflow templates) and on concrete workflow instances (executables). Development of a virtual laboratory faces some major challenges, however, including the highly distributed and heterogeneous nature of virological, immunological, clinical, and experimental data; the high dimensionality and complexity of the genetic and patient data; and the inaccessibility and (lack of) interoperability of advanced modeling, simulation, and analyses tools.
Recent advances in grid computing tackle these problems by virtualizing the resources (data, instruments, compute nodes, tools, and users) and making them transparently available.
They talk about workflows as a systems-science language:
A custom-built approach isn’t sufficient for increasingly complex applications. Service-based distributed applications are ideal for automating and generalizing scientific workflows. Researchers can use them to combine data integration, analysis, and visualization steps into larger, automated "knowledge discovery pipelines" and "grid workflows."
I am a student at the University of Michigan School of Information, where I rub elbows with many of the pioneers and leading practitioners of CSCW and HCI - Gary and Judy Olson, Tom Finholt, Stephanie Teasley, George Furnas, and Suresh Bhavnani, to name a few. One of SI's research achievements in the course of the early research on scientific collaboratories is the identification of three factors that are requisite to the success of a large-scale distributed CSCW effort: collaboration readiness, collaboration infrastructure readiness, and collaboration technology readiness.
To apply these in reverse order to Virolab: one assumes in spite of the lack of detail in the article that the technology issues are solved, because there are a plethora of technologies, commercial and Open Source, that could be integrated, and that in any case demonstrate that the problem is solvable. Infrastructure should be no issue, given the state of the art in grid technology, which extends the prospect of a true "utility computing" environment with abundant computation, storage, and data transmission resources made available. But collaboration readiness? Ah, there's the rub. Are we willing to break out of our silos, to share data and specimens, not to mention the glory of discovery? Are we ready for "team science"? That, as I see it, is the real question. Virolab is just one data point in that particular experiment.
Comments