With a huge selection of off-the-shelf expression analysis software on the market, Nîmes, France-based biopharmaceutical firm CliniGenetics had plenty of bioinformatics infrastructure options when it launched just over two years ago. So which commercial platform did it standardize on? None of them.
The company is only 17 people strong, but founder and CSO Gérard Marguerie said there was only one way it could integrate its bioinformatics system with its research in cardiovascular target discovery: Build one from scratch. So with financial support from the French Ministry of Finance and Industry, CliniGenetics partnered with French IT firm CSSI to create the cardiovascular informatics system of its dreams, which it calls ALIVE.
CliniGenetics required proprietary software, said Marguerie, because it generates all of its own data from its in-house animal models. “The bioinformatics for the company is based on home-made data and not data mining,” he said.
In building the platform, the company not only avoided commercially available tools, but eschewed publicly available algorithms and software packages as well, opting to write every line of code itself. Added Nora Benhabilès, director of bioinformatics, “We need our own algorithms to analyze our own data.”
Now fully operational, the ALIVE system integrates information from reference pig models that are phenotyped for cardiovascular disease with biochemical data and information from microarray experiments on the Agilent chip platform. It comprises two main components: Express Profiler, which manages microarray image data; and Vascular Resolver, which analyzes the data to identify genes of interest in vascular cells.
Ten developers from CliniGenetics and CSSI worked for the last two and a half years putting the system in place, Marguerie said, and the work is already paying off — the integrated system helped recently identify three new metabolic pathways associated with the development of atherosclerotic plaques.
Express Profiler performs many of the same functions as commercial systems like Rosetta Resolver, Marguerie said — including image capture, spot segmentation, detection of the background, ratio determination, normalization, and clustering. CliniGenetics evaluated commercially available systems, including Resolver, but found that they were “incomplete and very expensive,” Marguerie said. In particular, he noted, the image capture step in most commercial systems was not up to the company’s standards, so CliniGenetics added a feature to its own software that performs segmentation and statistical analysis on each spot. In addition, the company found the normalization lacking in commercially available tools, so it based its normalization methods on “universal controls” generated from its animal models, which also eases integration with downstream analysis steps, Marguerie said.
This analysis is performed within the Vascular Resolver module, which is a database and a set of proprietary analytical algorithms for clustering genes according to intensities, target, time, phenotype, and other identifiers. Additional statistical methods are included in Vascular Resolver to link expression data with other biological information such as kinetic or phenotypic data.
While ALIVE was designed for the company’s cardiovascular research, Marguerie said it could easily be duplicated for other disease areas. CliniGenetics is considering making the system available to other researchers, he said, but was quick to add, “We’re not a bioinformatics company.” If it can find additional partners for further development of the system — while remaining focused on its primary goal of developing drug candidates — look for a publicly available version of ALIVE before the end of the year.