As many commentators have noted, there is less than meets the eye in the content of recent statements by President Bill Clinton and Prime Minister Tony Blair that the raw data from human genome projects should be released rapidly into the public domain. But they affect the climate in which decisions over patents and investment are taken. Now the presidents of the National Academy of Sciences and of the Royal Society have issued a joint statement in this week's issue (see page 325) urging that the human genome sequence “must be freely available to all humankind”, and that patents should only be awarded to applications in which significant utility is demonstrated.

The principal, if unmentioned, target of these concerns is Celera Genomics, the private rival of the publicly funded human genome project. The existence of the public project is a guarantee against a potentially catastrophic monopoly situation. Celera's financial interests lie in its combination of intellectual property derived from genes and their variability, and from charging subscribers for access to software that accurately pulls out genes and makes good predictions of function by integrating information from a range of sources, including sequence homologies in different databases and relevant literature. The value of such services is reflected in the $283 million acquisition by Celera of the software company Paracel.

But what are Celera's terms of access? In statements to Nature, the company has promised that academic researchers will have free access at the time of publication — some months after subscribers get unconditional access — to the primary sequence data from its website, or on DVD, without any restriction, and can publish and seek patents on discoveries without any 'reach through' rights (see page 324). Anyone, except competitors, will be able, for example, to use the data to develop DNA chips, it claims.

The sticking point is annotation — the identification of gene sequences and assigning probable function. Traditionally, researchers have annotated sequences and made the results available freely. But it appears that Celera would not allow researchers who are not subscribers to take the whole genome, annotate it and distribute it in competition. This restriction appears unreasonable, given that the challenge of making good predictions of function is in the software and not in the sequence itself.

Celera's ambitions highlight the potentially changing rules of the game of databases. There is every possibility that large data sets will increasingly be held by privately funded ventures.

Nature's policy is that human sequence data should be deposited in a reliable, publicly available, unrestricted and free database. GenBank and its equivalents in Europe and Japan satisfy those conditions. But there is no insistence in principle that deposition be in databases that are publicly funded — the conditions of access and the community's confidence in the long-term sustainability and accessibility of the database are what count. The value of databases such as GenBank lies partly in their role as an honest broker of information for the community at large, and any private owner of a database would do well to have earned a strong reputation in that respect too.