Proteome Explorer.
Roger Kramer, Cammie Lesser
Department of Medicine (Microbiology and Molecular Biology) MGH/HMS, 65 Landsdowne St., Rm 417 Cambridge, MA 02139, USA
As high-throughput methodologies proliferate in the systems biology community, whole-genome datasets are proliferating even faster. So many gigabytes fo data are now publicly available that one may imagine discoveries lay hidden in plain view. Software for data mining is also proliferating, but such tools often move rather than solve the fundamental problems posed by our abundance of riches. Algorithms cannot make discoveries in the strictest sense -- an algorithm presumes both the question and the form of its answers, yet they produce still more data in which discoveries may lie. Genuine discovery still requires the human brain's capacity for open-ended pattern detection which exceeds anything machine implementable, and graphics are the only means to communicate the quantities of dataa available efficiently to the brain. In a well-designed visualization a researcher may preceive subtle, unanticipated relationships to which an algorithm is blind because the researcher brings all their knowledge to bear whereas any given algorithm applies only the heuristics a priori deemed relevant and embodied in code. Software that creates such visualizations must be accessible to the people with the knowledge, first and foremost biologists. These issues motivated our lab to develop software we call Proteome Explorer (Px). While superficially similar to other currently-available network kvisualization tools, Px differentiates itself in its sismsplicity and emphasis on iterative, interacative exploration. The software engineering in Px is focused more on the user interface than on novel algorithms. The ease with which intelligible proteome visualizations are produced encourages experimentation with different juxtapositions of data. Px' visualizations are automatically "hot-linked" to public data repositories to simplify follow-up research. And Px provides a plug-in architecture to aaccommodate additional automation capabilities. Px is a tool to aid, not replace, the researcher's intuition and shorten the turn-around time between "fishing expeditions" and hypothesis-driven experiments.
Return to YGM 2004 Home at SGD