Solving genomics database problem could save years of research

Solutions to the Tower of Babel annotation problem that haunts genomics databases are showing they can shave "several years" off the drug discovery process according to a leading US bioinformatics worker.

Dr Dan Davison, associate director, bioinformatics and applied genomics, at Bristol-Myers Squibb's (BMS) Pharmaceutical Research Laboratory in New Jersey is in the forefront of those attempting to deal with the Tower of Babel issue.

It relates to the problem caused when different researchers christen the same proteins or gene sequences with a confusing jumble of distinct names.

Outlining the problem in a seminar at the University of Queensland's Institute for Molecular Bioscience, Davison said the same proteins and gene sequences can acquire a dozen or more labels as they are uncovered in different labs.

"Everybody that works on something calls it by a different name. For example, among the kinases involved in signal transduction in cancer, there is a kinase for which the convention is to call it MAP2K but it can be referred to by nine other names."

Causing even more confusion, the opposite problem also occurs in that "you can frequently have the same name for 20 different proteins," he said.

The result can impede the drug discovery process because the nomenclature clutter in genomic and proteomic databases creates false trails that lead researchers astray.

Davison and his co-workers at BMS have addressed the problem by creating an integrated solution called GeneTracker which among other things harmonises different annotations as they are imported from a wide range of external gene sequence and expression and proteomics databases.

"Using our tools with the combined sequence databases and combined expression and proteomics databases, we can save several years in the drug discovery process, he says.

"It allows us to identify liabilities sooner or to say we clearly have a proof-of-principle. If you have to do that (proof-of-principle) in a lab it takes forever but you may be able to derive from proteomics that information that says here is the single experiment you need to do to confirm it."

Such efforts can lop several years off the drug discovery segment of the 10 to 15 year period needed to bring a drug successfully to market, he said.

While BMS' has developed GeneTracker as a proprietary system, the Human Genome Organisation (HUGO) is driving harmonisation in a more generic way in Europe and progress is also being made in the US, Davison said.

"It involves a certain level of cooperation among people working in specific areas to agree on certain names."

Davison points out that for a researcher ferreting out new facets of a particular sequence, "it represents your standing in the international community and the recognition of your peers...it amounts to your baby."

"So one always wants things to go faster but at the same time it is reassuring that many of the experimentalists are willing to give up their baby (by not insisting on naming rights)."