DISCOVERY TOOLS · A Columbia LIMS named SLIMS picks old drug for new disease
BY MARK D. UEHLING
February 15, 2005 | Spinal muscular atrophy (SMA) is the type of disease you wish you'd never heard of. The industrial world's leading genetic disease killing infants under age two, SMA is a neurodegenerative condition that affects 1 in 6,000 live births. There is no treatment.
Columbia University's Brent Stockwell, with joint appointments in biological sciences and chemistry, had a few ideas about SMA. He had a human cervical cancer cell-based assay and a library of 47,000 compounds to test against it. There was just one problem. The high-throughput screening campaign he wanted to run, he knew, would be fraught with data glitches using any commercial laboratory information system (LIMS).
So Stockwell did what scientists have always done: He cobbled together his own tools, which he then offered to the world for free. Sounds great — unless you're selling a LIMS for the price of a house. But he says it had to be done. "There is nothing out there to allow academic researchers to do the type of high-throughput screening and high-throughput biology that we're starting to do," he says.
He did try several commercial LIMS: "We were quite disappointed that they would not do the type of analyses that we needed to do. They don't have the flexibility that one requires. When you go to industry researchers who do similar types of high-throughput screening, 90 percent will tell you that in the end they need to design their own LIMS because the commercial systems are really insufficient."
One Man, Two Years
Thus a chemist-programmer in Stockwell's lab, Brian Kelley, spent two years creating a new LIMS called the Small Laboratory Information System (SLIMS). In this case, "small" is relative: SLIMS can currently handle databases 3 gigabytes in size. Among other things, SLIMS can normalize lab values, correct the data, identify active compounds, and probe structure-activity relationships.
Often, Stockwell notes, the vagaries of high-throughput research mean that microfluidic equipment or plate-handling robots introduce systematic errors into the data. Clogged nozzles, temperature variations, humidity issues, evaporation — just about anything might cause all the wells at the edge of a plate to have a consistent difference from those at the center. Like breezes sweeping across a lake from different directions, such statistical patterns can be tricky to detect. In the past, Stockwell says, there was little that could be done: "We just had to live with the increased false-positive and false-negative rates."
Now, with SLIMS, the software can detect and correct errors using a Fourier transform algorithm. "In every plate," Stockwell says, "we assign the probability that the signal on that plate is random or is associated with a systematic pattern. For plates that have a systematic pattern, you have a choice of either using the data as is or correcting that pattern to eliminate it."
Another advantage of his SLIMS: judicious procrastination. The software permits scientists to postpone picking a data schema, and to change the experimental design in the middle of a project. Commercial LIMS vendors, Stockwell says, typically charge hefty fees for that. Their customers are locked in to their initial vision of the HTS campaign. "The idea with this system is that you don't have to make those decisions up front," Stockwell says. "You can continually change the type of data analysis or screening format you're doing."
Who Needs a Chemist?
STOCKWELL: He wanted a cross-platform tool
The SLIMS software is not just for lab slaves. It has already helped Stockwell and his Columbia colleagues identify indoprofen, a cousin to ibuprofen, in the treatment of SMA. (Indoprofen was approved as a painkiller in the United Kingdom in the 1970s but withdrawn because of minor gastrointestinal side effects. It's now off patent.) Using SLIMS, Stockwell and his collaborators found that indoprofen seems to help patients missing one of two key genes for spinal muscular atrophy.
Once two papers on SMA were published late in 2004, Stockwell posted the SLIMS application and tutorials at sourceforge.net. In academia, Stockwell thinks, SLIMS' cross-platform design (running on PCs, Macs, and Linux machines) should be appealing. "Many of the instruments run off of PCs," Stockwell notes. "We find an increasing number of PCs infiltrating the lab. It obviously took more work to develop it for all three platforms. But we felt it would increase the usability and probability that others in academia would adopt it."
Now, Stockwell says, his lab and collaborators at the National Institute of Neurological Disorders and Stroke will be trying to tweak indoprofen. They're exploring both a re-formulation to get the drug into the spine and across the blood-brain barrier — and a modification that produces a completely new molecule. The SLIMS software, fortunately, exports data into both Excel files and MDL's SD format for chemical structures. About the only thing the software can't do yet — allow more than one user to access the database — will be addressed in a future release.
For more information, Chem Biol 11(11), 1495-1503; 2004 and Chem Biol 11(11), 1489-93; 2004.
PHOTO BY KATHLEEN DOOHER