Those charged with hearing pitches from software vendors who want to sell wares to biotechnology companies don't like these words: "enterprise-wide solution." They don't want to hear the generic wonders of the "solution" being pitched, they don't want to hear marketing buzzwords or that the software will revolutionize the pharmaceutical business. They won't believe that kind of approach and they will show the software vendor the door, perhaps within minutes, without an invitation back.
So said James Golden, the business development manager at 454 Corp., which focuses on developing technologies for massive-scale genome analysis, during a panel discussion titled "How Do You Measure the ROI for Informatics?" as part of the 8th Annual Drug Discovery Technology conference Tuesday in Boston. Besides Golden, the panel included representatives of pharmaceutical companies.
Vendors need to understand the "domain" of the companies they pitch products to. "I need you to clearly demonstrate to me in the first five minutes of our meeting that you understand what I do," said Golden, whose company is based in Branford, Connecticut. "You'll get a second meeting if you can do that." And, he said, be specific in demonstrating knowledge about the potential customer company.
His suggestions were in response to a question about whether companies like his and the pharmaceutical vendors prefer to buy commercial software or build their own. Often, such companies have to build their own because they can't find commercial software that fits their needs, the panelists said. However, there is a preference for buying what is available when the tools are suitable.
Biotech companies tend to have the view that what they do is "totally unique compared to everyone else" and so software tools have to be developed in-house, he said, adding, "I'm not so sure that's entirely true."
Vertex Pharmaceuticals Inc., with headquarters in Cambridge, Massachusetts, buys a lot of software, but much of it is "widgets," said Mark Murcko, the company's vice president and chief technology officer. "We're finding more and more we're buying widgets rather than systems," he said.
That's because commercial applications are generally not well tested in real-world scenarios, so score poor marks in both validation and applicability, Murcko said. For instance, a commercial search engine might take three minutes to scour the massive amount of data a pharmaceutical company has at its disposal, while a homegrown engine employs an algorithm that takes three minutes to go through the same data.
It's the old complexity issue, which isn't easy to deal with, said Richard Roberts, the worldwide head of research informatics at Pfizer Inc., which knows a thing or two about massive amounts of data.
But there's another big issue when it comes to whether Pfizer, with headquarters in New York, will decide to buy commercial software, and that's licenses. The oft-used method of charging per user needs to go, he said, with a model based on actual usage taking its place. "The vendors that do that will have a very strategic relationship to play within Pfizer in the next few years," Roberts said. "Those who don't, won't."
Various other factors also come into play when those in charge of technology at biotech companies try to figure out if they're getting their money's worth. Different groups of people within such companies have different needs and those needs may change often, and that's a particular challenge when a company operates from multiple locations, said John Hill, executive director of drug discovery and exploratory informatics at Bristol-Myers Squibb Co., with world headquarters in New York.
The technological innovation his company most needs in the next five years involves data and software integration, Hill said, "and how to make information available to the right people at the right time, in the right context."
Predictably, integration came up repeatedly as a major challenge and as an area needing attention in the near term.
Roberts added to that the need for informatics to help scientists better understand the genome. Now that they know what the genes are, they need to figure out what they do and how they work. Also important to pharmaceuticals is predictive toxicology, with good models developed that help companies figure out early in the process of discovering new drugs whether the drugs are dangerous for humans to use. Good predictive toxicology technologies would more than double the productivity of pharmaceutical companies, he said.
But perhaps the largest issue has little to do with technology, but with getting recalcitrant scientists excited about tools available to them. The key to that is to let the scientists be scientists and to not try to push them to be interested in topics that just don't wow them, various panelists suggested, perhaps to the chagrin of engineers in the audience who advocated for big pharma to hire some engineers for upper management to make decisions from that point of view. (Such advocacy came up elsewhere at the conference as well, indicating the historical friction between science and engineering and creating a stir among audience members and panelists alike.)
"Scientists don't ask questions like, 'Should this pull-down menu be next to that pull-down menu'," Murcko said. "Scientists don't care about those kinds of questions."