LONDON - The bioinformatics company Inpharmatica Ltd. has launched the first comprehensive database of pre-calculated structure-function relationships. This brings together all the information on sequence, structure and function relationship for the half-million or so gene products in the public domain.

The database, called Biopendium, currently comprises over 100 million such relationships.

Mark Swindells, scientific development director, told BioWorld International, "Biopendium will be immensely more powerful than existing methods of finding relationships. At the moment anyone with a protein sequence looking for relationships with proteins in public databases is forced to go through half a million sequences. We have bitten the bullet and calculated everything against everything, allowing users to compare their protein sequences directly to identify interesting relationships."

Inpharmatica, based in London, said it is in discussions with three major pharmaceutical companies that are interested in using Biopendium. The company will offer services such as the incorporation of proprietary gene sequences, and client-specific protein structures.

The amount of information in public databases is doubling every 18 months. Swindells said Inpharmatica is currently recalculating the relationships every three months, but has recruited more staff so it can update the database every month.

Apart from speeding the process of validation and lead optimization, Swindells said Biopendium would reduce the need for bioinformaticists. "The real thing is that this provides answers. A lot of people sell software tools, but they are a bit like Formula One racing cars - unless you are a bioinformaticist, you can't use them."

Molecular biologists want to be able to find answers for themselves and to get them quickly. For example, using Biopendium, it would be far quicker to find a target for, say, a new antibiotic. "Many bacteria are completely sequenced, but the question for someone attempting to design an antibiotic is, 'Which of the 4,000 genes do I choose?'"

Biopendium would be able to identify all the sequences in a particular bacteria that did not occur in humans, and were enzymes. "This brings 4,000 down to around 100; in other words, rapidly eliminating the targets that are no good."

Users also may find it possible to resurrect previous projects. "There are lots of dead projects around, which found inhibitors to targets that turned out not to be good targets. Biopendium can find proteins that are similar, and therefore likely to bind molecules that have been developed already," Swindells said.