What’s Genedata’s New Year’s resolution for 2004? According to CEO Othmar Pfannes, “This year we plan to be a little bit more outspoken in communicating that we’re not just a pure software provider, but we actually do quite exciting drug discovery process work.”
The Basel, Switzerland-based firm has already made good on that goal, with the publication of a paper in January’s Genome Research describing results from a bioinformatics collaboration with Bayer in the area of antimicrobial drug discovery. Genedata has worked with Bayer since 1998 in a fruitful partnership that Pfannes admitted is a bit of a secret in the industry. “It’s actually not our style to publish this type of thing, but we are actually doing quite a lot of exciting research, not only with Bayer, but with a few others,” he said. Now, he said, it’s time to “make a little bit more noise because I think we can help other companies too.”
Genedata is known for its shrink-wrapped Expressionist and Phylosopher software products, but Pfannes said that deals similar to the one with Bayer — with companies such as Altana, Novartis, Schering, and AstraZeneca — currently make up about 30 percent of its revenues. Genedata’s collaborations are based on three major project areas: target discovery, lead discovery, and — just within the last year — toxicogenomics. Pfannes said that he foresees the company’s collaboration business growing as it continues to build its strengths in these areas.
But the principal benefit of pharmaceutical partnerships for Genedata, according to Pfannes, is the opportunity they provide for improving its software. The company signs only non-exclusive agreements, he noted, so it is free to embed any new technology in future releases of its products. Some capabilities developed for the recently published Bayer study, for example, are already in current versions of the company’s software, and the Phylosopher product itself grew out of an earlier research project with Bayer. “I would say such a project adds about 10-15 percent new functionality to the software,” Pfannes said.
Bayer doesn’t seem to mind sharing the wealth. The company recently extended its collaboration agreement with Genedata for another three years, according to Pfannes.
Putting the Software to Work
In the Bayer antimicrobial project, Genedata provided the bioinformatics expertise necessary to cull through thousands of microarray experiments and derive a new approach to identifying novel drug candidates that are based on promoters in biological pathways. Essentially, the approach eliminates the need to identify specific targets for screening by using so-called promoter-inducible assays — assays that use cells with a reporter, such as the firefly luciferase gene, fused to a promoter in a known pathway. After screening, the presence or absence of luminescence indicates whether a compound is perturbing the pathway of interest, even if the actual inhibited target is unknown. Such reporter assays offer promise in screening for compounds with a defined mechanism of action, but their use has been limited so far due to the small number of known promoters.
According to the Bayer and Genedata collaborators, the key to identifying new promoters lies in microarray data, and lots of it. In the Genome Research paper [14:90-98 (2004)], they describe how they used expression data as a starting point in identifying a novel transcription factor binding site (TFBS) motif that would have been lost in the noise if they had looked at sequence data alone. They started by building a “reference compendium” of expression profiles for nearly all 4,100 Bacillus subtilis genes treated with eight different antibiotics. Focusing on a key gene in the B. subtilis fatty acid biosynthesis (FAS) pathway — yjaX — the researchers then used Genedata’s Expressionist software to mine the data in the compendium to find a set of 20 genes with expression activity strongly correlated with that of yjaX. Mapping those genes onto the B. subtilis genome to identify highly correlated genes located next to each other and transcribed in the same direction narrowed the set down to 10 regions in which to search for regulatory motifs using pattern-recognition algorithms in Genedata’s Phylosopher software.
The software did identify a pathway-specific TFBS motif, but it would not have been possible without the compendium of expression data, according to the researchers. “[I]t has to be emphasized that the (annotation-based) knowledge of all fatty acid biosynthesis relevant genes in B. subtilis, of which all upstream regions have also been compared, did not enable us to identify the putative TFBS motif … Only the expression data revealed the genes that are transcription-correlated with yjaX, and allowed the reliable prediction of the corresponding transcription control regions. Thus, the results of this work are critically dependent on the expression profiling data that were necessary to restrict the number of candidate regulatory regions,” they wrote.
The team then constructed an assay based on an engineered reporter strain of B. subtilis in which the luciferase gene was fused to one of the promoters that contained the TFBS motif. The assay correctly identified two known FAS inhibitors, while non-FAS inhibitors did not induce the reporter system. “It should be emphasized that from a drug discovery standpoint, knowledge of the regulatory protein is not required” for the approach, the authors noted.
For Bayer, the results of the study are promising for future drug discovery efforts. Based on the results of the study, “inducible-promoter assay technologies are likely to become a widely used tool for rational drug design campaigns aiming at the identification and development of novel antibiotic agents,” the researchers wrote, adding that they have already used the method in other research projects that have not been published. Specific opportunities, they note, lie in pathways “that have been recognized to harbor promising, but as yet unex-ploited targets,” and in revisiting targets that failed to provide promising lead compounds in previous target-based screening projects.
For Genedata, the study also serves as validation for its technol-ogy and the company’s value as a research partner. Pfannes cited the promoter identification aspect of the project — coupled with the large-scale microarray analysis piece — as particular points of pride for the company, and predicted that more research teams will be able to adopt similar approaches as the costs of microarrays continue to fall. “I remember when we first talked about this project, and we said we need at least 100 experiments, and they said, ‘Oh my God, what an expense!’ Now I think it’s not so much of an issue anymore … Microarrays are now at a price where you can do 1,000 microarrays — it’s not a multi-million dollar thing anymore.”
That, of course, leaves more in the budget to pay for software, and a little bit of extra help putting it to work, if necessary.