Drug Industry Pegs Pitfalls of Keeping Data ProprietaryBy M.L. Baker | Posted 06-23-2005
Scientists today can generate data in a single afternoon that previously would have required years of work and been considered worthy of a doctoral thesis. But no one seems to be using this data well, largely because neither intellectual property law, information systems nor scientific training allow researchers to wrest valuable secrets from these masses of data.
"We can generate huge amounts of data, far more than we can interpret," said Duncan McHale, director of clinical pharmacogenomics at Pfizer Global R&D, who spoke at a panel on industry-academia relationships.
But McHale said companies hold onto the data as proprietary rather than share it in a way that might make trials faster and avoid years on drugs making more and more mistakes. The drug industry is reinventing a lot of wheels, and some expensive ones at that.
The industry needs to find a way to make data "precompetitive," he said. Drug executives must be trained to find ways to share data so that any common gain that comes from data mining is not seen as a private loss.
McHale said drug executives worry that they could be penalized with patent licenses for the very data they generate if they put it into a common repository that other companies use to generate predictive tests or novel tools. The fear comes out of patenting activity in the early days of genetic sequencing, when researchers would put intellectual property over a series of genetic letters.
Technology required sequencing genes in pieces, and researchers would file patents on pieces, without a clear idea of what the gene did or how knowledge of its sequence could be used to make drugs. Even today, researchers are not always certain of when they are infringing on another's patent.
McHall said he can imagine researchers poring through data, automatically generating measurements that might predict drug response or susceptibility to disease, and then patenting the use of those measurements.
For this reason, "People should not be able to dive into a database and pull out a patent," McHall said. Instead, people who obtain patents should do confirmatory experiments to demonstrate that a tool is robust.
But even as drug industry executives called on mathematicians and computer scientists to join their ranks, a conceptual divide was apparent.
At a separate panel, information scientists said they have no interest in what's being called "wet lab work" and clinical trials. Instead, they want to create tools that can probe the data in silico. They dream that experiments in cells, mice and people one day all will be confirmatory. The actual exploration of ideas will come from virtual models and data mining. Where the data would come from was unclear.
A potential model is that information scientists would generate hypotheses based on data from drug companies. While the IT people would own the software, the drug companies would own the actual findings.
But drug researchers tend to think that proprietary software is not necessary, and that open-source tools will suffice. Not surprisingly, information scientists disagree. Without the proper IT systems, "You [drug researchers] spend a lot of time rediscovering what you already have in your archives," said M. Vidayasagar of Tata Consultancy Group, which comprises IT specialists with a $90 million service in life sciences and health care.
Instead, software engineers need to review all of the processes in the drug research process and anticipate what software is needed.
For example, Vidayasagar said, databases that use static fields must come to an end. Nowadays, everyone wants to analyze patient data by BMI, a ratio of sorts between height and weight. Previous patient data did not consider that, he said, even though that information usually existssomewhereon file.
The best entity to sort out who can use what data for what purposes is probably the federal government, one moderator said. But companies will certainly refuse to release their data for trials they fund. To get access to such information, the government might find itself paying for more clinical trials.
Check out eWEEK.com's for the latest news, views and analysis of technology's impact on health care.