More exacting extracting

SAN FRANCISCO—Reel Two Inc., and Advanced ChemistryDevelopment, Inc., (ACD/Labs) have established a new collaboration that theybelieve will result in more accurate conversion of chemical names extractedfrom patent texts into chemical structures.

Under the open-ended collaboration, Reel Two hasincorporated ACD/Labs' ACD/Name to Structure Batch nomenclature software intoits SureChem portal and will offer it as an option with SureChem Enterprise."What SureChem does is it indexes the full text of all major patent datasources for all of the chemical names in the text and converts them, wherepossible, to structures," says Nicko Goncharoff, senior vice president at ReelTwo.

SureChem, which can be searched for structures or keywords,developed out of demand among drug discovery researchers for a full-textautomated product: existing products were good, but not comprehensive, saysGoncharoff. Patent databases contain over 16 million compound names and,Goncharoff adds, "customers wanted a product that could be used on their owndata as well as on existing collections of patent data." SureChem Enterpriseindexes and integrates companies' proprietary data on small molecules, allowingthem to be run against world patent data.

Antony Williams, vice president and chief science officer atACD/Labs, says that even unambiguous names can cause conversion difficulties."Our focus has always been on the quality of names and structures that we generate,so in this particular project, our intention was to make sure that we wouldproduce the unambiguous structures." Williams says the collaboration will giveACD/Labs an opportunity to receive feedback on potential difficulties withconverting challenging names. He acknowledges inherent imperfections, saying"we are probably the best at doing it today, and we are far from perfect."

ACD/Labs and Reel Two expect to launch SureChem by earlyMarch, with the date dependent on how quickly 300 gigabytes of patent data canbe processed, says Goncharoff. Portal launch will precede enterprise sales.Although Goncharoff sees SureChem Enterprise as a "top-tier solution" for bigpharma because of its price point, both he and Williams see value in marketingthe portal to other clients, including academic researchers and biotechs thatwork with small molecules.

Reel Two had not finalized pricing at press time, butGoncharoff expects one-year, single-user portal subscriptions to cost around$3,600, with enterprise software running roughly $50,000 to $75,000 annually,including licensing of name-to-structure conversion software, such as ACD/Labs'Name to Structure. Thirty-day portal trials are also available, saysGoncharoff.

The collaboration itself arose after Reel Two heard positivefeedback from customers about the precision of ACD/Labs' structure conversiontools. The companies have a straight licensing agreement for the Web siteversion of SureChem and will take enterprise software sales on a case-by-casebasis because enterprise deals can be complicated. Combining two of the best approaches in the industry,says Williams, has resulted in a synergistic and smooth relationship. "If wesee a success it will continue," he says.