Scientists take a closer look at ELNs

Atrium recently surveyed 460 user scientists about resistance to an ELN implementation within their company, and only 5 percent said everyone in their department was eager to use it. Thirty-five percent said half the people resisted the change, and 10 percent said the majority in their company resisted.

In the field

There's widespread agreement that the most successful implementations are bottom-up, not top-down--that is, the interest has to be generated and sustained by the chemists themselves.

That proved to be the case at Arena Pharmaceuticals, where Philip Skinner is both a chemist and the project leader on the implementation of software from vendor CambridgeSoft. He and a few other bench chemists oversaw the day-to-day implementation.

"We understand the software and the work processes, so we can tailor the notebook to our needs," Skinner says. "The IT group supports it, but they are not dominant in the project. The users have brought it in for themselves."

Rolled out in March 2005, the ELN is now used by 50 chemists at Arena, which is developing drugs to treat metabolic, cardiovascular, inflammatory and central nervous system diseases. Although they have a fully electronic archive and use digital signatures, the researchers currently use a "hybrid system" in which they also print out all their work as well. "That's the only onerous part," Skinner says. "It's a matter of precedent. Attorneys want to see precedent of electronic signatures as valid."

As one of the pioneering ELN customers, Array BioPharma of Boulder, Colo., helped CambridgeSoft shape its commercial product more than five years ago, says James Rizzi, director of computational technology at Array. By working with a small group of five enthusiastic users on several iterations of product development, they learned some lessons the hard way before rolling a product out to approximately 100 users.

Acceptance was slow at first. "There's resistance to typing everything, resistance from slow changers who say they are too busy to learn it," Rizzi says. There are also "Big Brother" fears that management will use it to measure the amount of output of each scientist.

To overcome those concerns, Rizzi stresses, "You have to prove to the users its advantages on a tactical level. You have to show how it makes some things easier for them." For instance, after a machine that does product validations automatically saved its results into the electronic notebook, the efficiency gains made obvious sense to the scientists.

"We also found it valuable to put computers in the labs with them, so they didn't have to return to their desks 30 to 50 yards away to enter information," he says. "They could just turn around and incorporate observations immediately." At Cambridge, Mass.-based Millennium Pharmaceuticals, what started as looking for a way for chemists to access and search chemical structure information grew into a larger ELN project.

"We started out wanting to give them a database to store and retrieve reactions to save them time and effort," explains David Sedlock, director of research informatics. Those features just happened to be found in larger ELN products that also had a mechanism to work with the legal team for patent preparation purposes.

In June 2004, Millennium began a pilot project with an ELN from Synthematix, which has since been purchased by Symyx Software. After working out customization issues, the project team rolled it out to approximately 100 chemists in June 2005.

"We built middleware that connects the Synthematix ELN to other applications the chemists use, so they can access that data without having to leave the ELN environment," Sedlock says. Early on there's little data in the database to retrieve, so the value of the ELN becomes more apparent to the scientists over time.

Sedlock's biopharmaceutical company doesn't have the resources to manage a cost-savings analysis, but he says that the goal wasn't to save time, but to increase the capability of chemists to easily retrieve information. "The biggest value will be to look back two to three years from now as the knowledgebase is built up," Sedlock says. "We could then look at the reactions stored and measure how it's been used."