Brent Stockwell’s lab at Columbia University learned the hard way what many pharmaceutical research groups have known for years: “There is no commercial [LIMS] system for high-throughput screening data, and that you always in the end need to build your own.”
The difference, however, between those industry groups and Stockwell’s is that the fruits of his lab’s labors will be available for other research teams facing the same problem. The Columbia researchers have recently released their home-grown HTS LIMS, called SLIMS (for Small Laboratory Informatics Management System), on Sourceforge (http://slims.sourceforge.net/), and published a paper outlining the system in the November issue of Chemistry & Biology.
“We really did not want to develop our own system in the beginning,” Stockwell told BioInform. “We wanted to simply purchase a commercial system that would allow us to collect and analyze data associated with chemical genetics screens.”
Stockwell’s team did procure a LIMS system — valued at $100,000 — from an undisclosed vendor, and soon found “that it was simply not compatible with the kind of work that we were doing.”
Stockwell chalked up the system’s failings to the inherent differences between pharmaceutical labs — for which most commercial LIMS are designed — and academic groups like his own, which are beginning to adopt high-throughput screening technologies to better elucidate biological mechanisms, such as pathways and networks.
“In industry, often there’s a separation between chemists, biologists, and automation specialists … but in the kind of research that we do, it’s continuous collaboration and data exchange between chemists, biologists, and automation specialists,” Stockwell said. “So we needed a system that allowed the rapid and continuous data transfer between those groups.”
In addition, he said, commercial LIMS tend to be inflexible. “They’re built toward doing one assay over and over and over again in exactly the same format, whereas what we tend to do is … develop an assay, then we start doing a screen with that assay, and then we might change the assay as we go because we start to find problems that weren’t apparent when we were doing a smaller number of experiments,” he said. “You can’t change in mid-stream what you’re doing with the commercial systems, but you can do that with SLIMS.”
SLIMS includes a cheminformatics engine for substructure searching and similarity searching, a database that can be shared between collaborators, and modules for plotting and analyzing assay data via self-organizing maps and other analysis tools.
Other features include a tool that automatically identifies and corrects systematic errors introduced in the experimental process, as well as an algorithm that searches the scientific literature to annotate small molecules with information about their mechanism of action.
“That’s the kind of thing that we couldn’t incorporate into a commercial system, but we’re able to build that right into SLIMS,” Stockwell said.
Stockwell’s team recently used SLIMS to help determine that indoprofen, a nonsteroidal anti-inflammatory drug, may increase the production of the SMN2-luciferase reporter protein, which is lacking in patients with spinal muscular atrophy. The researchers screened more than 47,000 compounds in the study, and analyzed the data with SLIMS. Stockwell’s lab published the results of this study in a second article in the November issue of Chemistry & Biology.
SLIMS is compatible with “any plate-reader-generated file,” Stockwell said, but it does not yet work with other types of high-throughput assay readouts, such as imaging-based high-content screening experiments. “We would in the future like to incorporate that kind of compatibility, but we haven’t done that yet,” he said.
The lab is currently expanding SLIMS to serve as an inventory system for lab supplies, and Stockwell said that his team is also working to incorporate data from RNAi-based screens and cDNA-based screens. They are also developing a web-based interface for the system.
Stockwell said that SLIMS is freely available for academic researchers, but that the licensing terms for commercial users “haven’t been decided at the moment.”