3.1 Scientific workflows  (Page 4/4)

This example shows that conducting this kind of data-driven approach to analysing complex biological data at the level of biological pathways can provide detailed information of the molecular processes contributing to the expression of these traits. The success of this work was primarily in data integration and the ability of the workflow to process large amounts of data in a consistent and automated fashion.

Workflow reuse

Workflows not only provide a description of the analysis being performed, but also serve as a permanent record of the experiment when coupled with the results and provenance of workflow runs. Researchers can verify past results by re-running the workflow or by exploring the intermediate results from past invocations. The same workflow can also be used with new data or modified and reused for further investigations.

The ability to reuse workflows and to automatically record provenance of workflow runs gives workflow management systems a large advantage over manual analysis methods and scripting. Manual analysis techniques are inherently difficult to replicate and are compounded by poor documentation. An example is the wide-spread use of ‘link integration’ in bioinformatics (Stein 2003). This process, of hyper-linking through any number of data resources, further exacerbates the problem of capturing the methods used for obtaining in silico results where it is often difficult to identify the essential data in the chain of hyper-linked resources.

Workflow reuse is also an important area within the sciences, and provides a mechanism for sharing methodologies and analysis protocols. As a result, repositories for finding and sharing workflows are emerging. One such resource, myExperiment , developed in collaboration between the Universities of Manchester and Southampton, provides a workflow repository and a collaborative social networking environment to support the in silico experimental process, and to enable scientists to connect with those with similar interests. The workflows dicussed in the trypanosomiasis use-case study are available on myExperiment , as part of a workflow pack. Many of these have already been reused in other studies. One such example includes the re-purposing of the microarray gene expression workflow to analyse gene expression data from E. Coli . This workflow appends a further workflow to include a means of information retrieval for future text mining applications (shown in Figure 1).

Discussion

Manually processing and examining results in biology is no longer feasible for many scientists. Data is dynamic, distributed, and often very large. This will not change in the near future.

The integration and interoperation of data between different and distributed resources is a vital part of almost all experiments. With the exception of a few supercomputing centres, most institutions do not have the storage, computational, or curation facilities to consider integrating resources locally. The ability to access and utilise many different resources from all over the world is consequently a large advantage of workflow technologies. It allows scientists to access computing resources far beyond the power available through their own desktop machines.

Building workflows is a practical solution to problems involving access to data and applications, but care still needs to be taken to exploit these advantages. Interoperation without integration may lead to unmanageable results which are difficult to analyse. In this event, the problem has not been solved, but simply transferred further downstream. Considering how results will be used and who will be analysing them is important. For example, designing workflows to populate a data model, or to feed into external visualization software, could reduce these problems. The provenance traces of the workflow runs can also help scientists to explore their results.

Designing these ‘advanced’ workflows requires a significant amount of informatics knowledge that many laboratory researchers cannot be expected to have. They do, however, need to use tools and software to analyse their data. The introduction of workflow repositories, like myExperiment, provides the wider research communities with access to pre-configured, complex workflows. Researchers can re-use established analysis protocols by downloading and running them with their own data. In some circumstances, they can even run Taverna workflows through the myExperiment interface.

Increasingly, workflows are becoming applications that are hidden behind web pages like myExperiment, or other domain specific portals. Instead of stand-alone tools, they are becoming integral parts of virtual research environments, or e-Laboratories. Users may not necessarily know they are invoking workflows.

The use of workflows in research can reduce many problems associated with data distribution and size. In the post-genomic era of biology, for example, this is extremely important. Biomedical science is a multidisciplinary activity which can benefit from advances e-Science in equal measure to advances in laboratory techniques. Sharing workflows and in silico analysis methods, with tools like Taverna and myExperiment, can lead to significant contributions to research in this and other disciplines.

References

Altintas, I. et al. (2004). Kepler: an extensible system for design and execution of scientific workflows. Proceedings of the 16th International Conference on Scientific and Statistical Database Management

Fisher, P., Hedeler, C., Wolstencroft, K., Hulme, H., Noyes, H., Kemp, S., Stevens, R. and Brass, A. (2007). A systematic strategy for large-scale analysis of genotype phenotype correlations: identification of candidate genes involved in African trypanosomiasis. Nucleic Acids Resesearch , 35(16). pp. 5625-5633.

Hull, D., Wolstencroft, K., Stevens, R., Goble, C., Pocock, M., Li, P. and Oinn, T. (2006). Taverna: a tool for building and running workflows of services. Nucleic Acids Research , vol. 34, Web Server issue, W729-W732.

Stein, L. (2003). Integrating biological databases. Nat Rev Genet , 4(5). pp. 337-345.

Stevens, R. et al. (2004). Exploring Williams-Beuren syndrome using myGrid. Bioinformatics , 20 Suppl 1

Stevens, R. et al. (2008). Traversing the bioinformatics landscape. W. Dubitzky (ed.) Data Mining Techniques in Grid Computing Environments . John Wiley and Sons. pp. 141-164.

Taylor, I. et al. (2003). Triana Applications within Grid Computing and Peer to Peer Environments. Journal of Grid Computing , 1(2). pp. 199-217.