Abstract
Functional information is lacking for many of the hypothetical proteins encoded within sequenced genomes. Richard Roberts proposes that a community-based approach might offer an effecient way to fill the function gap.