Confirmant set to release Protein Atlas

US company Confirmant is racing to finish up the first commercial version of its Protein Atlas for June, but life science researchers and industry analysts say a warm reception is far from assured.

Networking and wireless communications company Marconi and biotech Oxford GlycoSciences announced the Confirmant joint venture - and the Protein Atlas project - last June. Confirmant had a public "unveiling" of the atlas recently, at the Genome Tri-Conference in Santa Clara, California.

Confirmant reported that it had 7000 experimentally derived, protein-encoding genes in the atlas by February, and aims to have 10,000 by June, and about 30,000 in 2003. The atlas is a database of information on protein-encoding genes that includes detailed data on gene structure and the protein variants they encode.

The atlas promises to make the drug discovery process more efficient by providing large pharmaceutical companies and bio-tech firms with accurate information to identify, prioritise, and validate drug targets, say Confirmant officials.

"The whole genomics revolution didn't deliver," says Jonathan Sheldon, Confirmant's CTO. "It's actually the proteins which are the drug targets. Most of the protein databases out there are largely hypothetical; they're not based on experimentally derived data but based on computational predictions and computational algorithms, and those algorithms have many documented flaws."

Working backwards

Unlike the process that went into sequencing the human genome, accomplished by the Public Genome Consortium and Celera Genomics Group, Confirmant starts with proteins and works backward to the genes.

The company takes samples from cell lines, tissues and body fluids from different parts of the body and uses a combination of protein separation techniques including 2-D gel electrophoresis, affinity chromatography and subcellular fractionation. After experimentally deriving the amino code sequence that makes up the proteins, Confirmant maps the proteins back to the genome using mass spectroscopy.

Researchers hesitate to embrace the atlas before they examine the data, however.

"It all depends on what information is provided," says Sam Hanash, a professor at the University of Michigan School of Medicine and inaugural president of the Human Proteome Organization (HUPO). "If they are all involved in cancer, for example, then that would be highly valuable. It all depends on the context."

The size of the initial data set may not be significant enough to add value to current target validation projects if it is not very focused, say Hanash and other researchers and industry insiders.

"The market for this is limited," says Winton Gibbons, a life science analyst at William Blair & Co. in Chicago. Though an experimentally derived protein database like the atlas may not currently be on the market on a subscription basis, there are biotech companies and large pharmaceutical companies actively engaged in building up their own protein databases, he says.

For example, Large Scale Biology Corp has developed its own protein index, Gibbons says, and Celera, though it is moving away from an information-vendor business model, is also doing its own protein research.

Drug targets

But some of the genes mapped by the atlas have never been identified before by predictive or indirect methods, says David Palmer, Confirmant's CEO. These new genes could provide novel drug targets, he says.

In addition, the accurate, detailed data supplied in the atlas could help companies validate existing targets, as well as do compound screening, microarray design, positional cloning, and association studies, Palmer adds.

The atlas "will allow large pharmaceuticals to prioritise targets so they can focus their efforts in the most efficient manner possible," he says.

HUPO's effort to develop public domain protein information is just getting started, and thus public domain data will not erode the Protein Atlas' worth, Palmer says, which is what happened to Celera's gene-sequencing information.

Confirmant plans to make the atlas raw information available in XML format to allow companies to integrate it with their ongoing projects, or with software analysis tools. Industry insiders estimate that the atlas, with the tools, would be priced at about $US2 million per year, and Palmer confirms the price could be that high.

The joint venture, however, has already had to alter its business model. One reason Marconi was interested in the venture was to offer database information on a hosted, application service provider model, Palmer says. However, companies have told Confirmant they weren't interested in ASP access.