-
There is one thing stronger than all the armies in the world, and that is an
idea whose time has come. -Victor Hugo
Dessy will talk about the evolution of computer use in the initial stages of research -- to record a scientist's notes and data, for 'mining' information in institutional databases, and for large scale testing of compounds during the advanced stages of new drug discovery. He will also explore mistaken predictions, the challenges of human adaptation, and technology on the horizon. His presentation, "Some call the world a dreary place" (Comp 30), is Monday, Aug. 23, at 10:30 a.m. in Convention Center room 225-227.
"It is a fascinating period," says Dessy. "In 25 years, we have gone from computers with 4K of memory -- which wouldn't hold today's screen savers -- to the ability to process huge volumes of data."
In the 1970s, computing centers were not addressing the needs of chemistry, he recalls. "Industrial scientists were champing at the 'byte' because they knew computer power was central to their jobs. When PCs were introduced, many scientists realized they were the solution for the lab. But most scientists didn't know how to use PCs, let alone how to interface with lab equipment." Between 1972 and 1992, Virginia Tech and the American Chemical Society provided week-long courses for 5,000 scientists, giving them hands-on experience, letting them hook up lab equipment themselves, showing them how to collect data. "It changed the climate, not only creating acceptance, but enhancing expectations for computer applications," says Dessy.
Now, instead of the needs of chemistry driving lab computing, it's the needs of biochemistry, says Dessy. In genomics, scientists are trying to identify the sequence of billions of base pairs in genes from the DNA of all manner of organisms, and are well along the way to automating the entire process. Combinatorial chemistry and highthrough-put screening, used in modern drug discovery processes, also involves sophisticated automation.
"Remember how we used to discover drugs?" asks Dessy. "Someone would stumble upon a natural product with some level of activity, and a chemist would try to make variations that were better -- more active, more directed to a target disease. Each variation would cost about $10,000 to create and you might have to create 10,000 compounds to find one that made it through all the tests and trials to become a marketable drug. It would take years.
"So, the pharmaceutical companies decided to try to use computers to help design drugs from scratch," says Dessy. This began with attempts to use computer graphics to visualize how drugs mated with their receptor sites in the body. "Combinatorial chemistry began in 1992 when industrial chemists said, 'Let's not worry about what drug shapes ought to be. Let's combine classes of drug precursors and test all possible combinations.' If you start with 10 each of three classes of starting materials for a drug , there are a thousand possible combinations, each a potential new pharmaceutical.. Researchers now use computers and robots to make all the possible combinations," Dessy explains.
"Existing equipment can make 10,000 compounds a week. You load in the chemicals and push a button."
Then, high put-through screening tests thousands of compounds against target diseases -- dozens of pipettes drop samples into test media contained in plates having 384 or 1536 wells. . A company typically creates two million compounds and targets five diseases. If it costs 50 cents to create and test each compound, that's $5 million. And it's a gamble. One company even calls their automated chemical storage unit a 'Haystack'" But the tests are founded in fundamental biochemistry, says Dessy. "If we are taking fiscal chances in a discovery process we couldn't do without the equipment, at least we are not wasting time that could save lives. Speeding up drug discovery is important business."
Meanwhile, in the lab
While the processing stage of drug discovery has become fully automated, use of
computers in the lab is still evolving, Dessy says.
Everybody is familiar with scientists' notebooks, which contain detailed notes on process and discovery, every page dated and signed. Companies are now migrating to electronic lab notes. The researcher types in what they are doing, includes output from instruments, drawings of apparatus, spread sheets, and relevant articles from digital libraries and web pages. "It's complex and free form. You have to sign it in such a way so you know who did what," says Dessy.
"Maybe you use a signature pad similar to UPS or a smart card. Paranoid companies might scan fingerprints.
"The beauty is, you can share information with your colleagues -- as compared to my notebook, for example, where few people can read my writing," he says. The electronic notebook is meant to be shared. It speeds up the process of discovery development. The electronic lab book also makes it easier to prove discovery dates, to protect a patent. It is easier to search through large amounts of data.
However, electronic notebooks have not totally replaced traditional data. "There are concerns: How do we know who entered the data or who changed it and why?" says Dessy. "While the Food and Drug Administration (FDA) is pushing firms to implement electronic record keeping as fast as possible, they are not necessarily advocating dispensing with original documents. The subtle question is "What is original data?"."
Pharmaceutical firms normally have to keep original data seven to 10 years, "but they actually keep it forever because they don't know what will happen," says Dessy. "If they are sued, for example, records may prove that a particular drug reaction, foreseen by an individual, was not shared with those who decided to produce the drug. Increasingly, the responsible individuals(s) as well as the company are being penalized.
Another reason original records are often kept is the present requirement to "certify and verify" that a piece of software has or hasn't impacted the data. "The FDA doesn't want the scientist to be able to delete any data," explains Dessy. Obviously, it's illegal to change scientific data. "You also have to keep test results, whether you like them or not." But off-the-shelf software and operating systems may allow deletions. As a result, computers with zero-administration ability are being created for lab use. Many functions that home PC users take for granted are removed, to assure the integrity of the data chain. This also saves company resources in an unexpected way. "We all have colleagues who provide computer support, even though that is not their job," says Dessy. "We may consider them a resource, but industry loses a lot of productivity that way. Some consulting houses feel that the total cost of ownership of a networked computer is $8-12,000 per year. Companies are trying to move administrative costs toward zero.
Failed predictions and lessons learned
* "We use to think that one day robots would do many things for us," says Dessy.
"For a period in the '80s, we foresaw human-type robots in the lab. A dozen
start-up companies worked on it. But it turned out that it was too complex to
program such robots. We've gone back to devices that move from point x to point
y, and up and down, like the automated pipettes."
* Speech input is coming back around for the second or third time, he says.
* "People said data format problems would disappear, but we are still having problems with compatibility," he says. "Electronic lab notebooks are not new. People started talking about them in 1985. It is just beginning to come to fruition because of better solutions to the format problem."
Things don't happen instantaneously," says Dessy -- who knows first hand. "Sometimes you have to wait until there is generation of students who have been exposed enough to new technology to be willing to adopt it. If you are too early with an idea, it may not sell. The users have to be prepared to accept it. If you tell people they have to change how they work, it had better be more efficient."
The future
* A lot of people are using computer systems for mining data -- to find in the
massive amounts of data available the core that is relevant to one's current
efforts or to recover experiments done in the past. As more historical and new
data goes online and better search engines evolve, this will become an important
resource.
* We are getting better at visualizing scientific information. We can create holographic images of molecules and even interact with them physically to feel the strength of the bond between a pharmaceutical agent and the receptor, for instance, if we try to pull them apart.
* There is experimentation with alternative presentations of information. A lot of what we have now is visual. Using sound rather than the visible allows the use of tone, volume, direction, phase, -- for finer distinctions among data elements. For example, I can barely see the different between octane and hexane in the infrared spectrum, but I can clearly hear the difference if they are presented musically."
* Automated translation into foreign languages is becoming common. (La traduction automatisée dans des langues étrangères devient commune.)
"This computerized world is by no means a dreary place," concludes Dessy.
Dr. Dessy, currently teaches an Honors Colloquium, "Internet Impact," dealing with the sociological, political and economic impacts of the WWW, and supports a homepage dealing with Web matters. Learn more at www.chem.vt.edu/chem-dept/dessy/honors/
For more information, contact Dr. Dessy at: 540-231- 5842 or rdessy@chemserver.chem.vt.edu