May 12, 2006
| Discovering new drugs can turn science into a Tower of Babel. “In drug design,” says Daniele Bemporad, a scientist in the molecular informatics group of Johnson & Johnson Pharmaceutical Research and Development (J&JPRD) in Beerse, Belgium, “you have scientists from different disciplines talking together. This is difficult, because biologists are familiar with some data, chemists have other data, and they all use their own software.” Surprisingly, one Johnson & Johnson scientist solved this communications crisis with a special pie.
When a pharmaceutical team gets together to decide which compounds to put forward as possible drugs, it must consider many factors. Does the compound bind to the right target molecule? Is the compound soluble in water? Is the compound quickly metabolized after administration? Assimilating the answers to these and other questions creates a hodgepodge of information that is difficult to interpret simultaneously. “Scientists are also dealing with enormous amounts of data, large numbers of molecules, each of which has many properties,” says Bemporad. “If you try to plot everything on a series of histograms with different legends, then it’s a big mess. Nobody understands a thing.”
Trevor Howe, head of molecular informatics at J&JPRD in Beerse, wanted some way to combine all the data — one visual image that describes all of a compound’s characteristics. To reach that objective, Howe invented VlaaiVis (pronounced “vlie vees”). In Dutch, “vlaai” describes a fruit-filled pie, and “vis” means fish. But at J&JPRD, VlaaiVis does not mean a fish pie. Instead, vis is shorthand for visualization — thus, VlaaiVis is a pie chart used for visualizing drug data. Bemporad says, “VlaaiVis is basically a communications tool.”
For a given compound, VlaaiVis represents that potential drug’s features with slices of pie. If the compound’s ability to bind the desired target, for example, is optimum, that piece of pie will be completely filled, or colored in; if that characteristic is far from optimum, only a portion of the slice will be colored; and if that feature has not been tested, that slice will be blank. Different characteristics make up all the slices of the pie.
Bemporad explains this piece of software quite simply: “It’s just a pie chart. If slices are filled to the edge, that is good. If they are not filled enough, that means bad. No filling meaning that the data have not been generated.” In short, “A VlaaiVis pie shows how well a compound might perform and also shows what work remains to be done.”
To develop this software, Howe turned to the IT staff at J&JPRD. The software simply takes the data from a compound, compares it to best-case values, and codes related measurements with some intuitive color. If a set of characteristics relates to cardio-toxicity, for example, the slices in that section of the pie might be red. “This software does not create data,” says Bemporad. “It does not even calculate anything. It does not even edit that data, because if that happens you would not know what has been done.” The result is an overall assessment of a compound’s potential as a drug. “You just need to see how many slices are filled close to the edge,” says Bemporad.