When the draft human genome sequence was published it was referred to as the blueprint. But, as with any design plan, the challenge is in establishing how the plan is executed. This task lies at the heart of functional genomics, and now Kamath et al. provide an important contribution to this field with their tour de force systematic RNAi analysis of the Caenorhabditis elegans genome. Not only do they uncover the function of scores of worm genes, but they also provide important information on the worm genome structure and its evolution.

To determine RNAi phenotypes for as many C. elegans genes as possible, the authors constructed a library of 16,757 bacterial strains (equivalent to 86% of all currently predicted worm ORFs), each of which expressed double-stranded RNA (dsRNA) corresponding to a single worm gene. When worms feed on these bacteria (Escherichia coli is their normal diet) the dsRNA is internalized and mediates sequence-specific knock down of the endogenous gene. One by one, the phenotypes were scored, and 10.3% of the analysed ORFs gave consistent phenotypes, which the authors grouped into three classes: nonviable (Nonv), growth defects (Gro) and viable post-embryonic phenotypes (Vpep). The first class contains many universal eukaryotic genes, for example, those that encode essential components of the basal cellular machinery. By contrast, most of the genes in the Vpep class probably represent animal-specific genes, and their products affect processes such as behaviour or body shape.

A closer look at the genomic distribution of genes in each class revealed that genes with similar functions tend to be co-localized in large domains of the genome and are co-transcribed. The size of these clusters, however, suggests that any large-scale transcriptional co-regulation must be mediated by a mechanism other than the previously described open-looped chromatin.

Another interesting finding concerns the X chromosome. The fact that Nonv genes are underrepresented on the X chromosome, whereas those that encode components of signalling pathways and transcription factors are overrepresented, suggests that very different selection pressures operate on genes on the sex chromosomes compared with genes on the autosomes.

Two other studies, in which the same approach was used to address more specific biological questions, have also been published. Ashrafi et al. used the same RNAi library to look for genes that regulate fat storage and mobilization. Among the 305 genes that reduce body fat stores and the 112 that increase them, they identified those genes that have mammalian homologues, some of which have already been implicated in fat metabolism. The worm fat-regulating genes fall into three main pathways (insulin, serotonin and tubby) confirming that fat metabolism is conserved in metazoans and that the worm can be used as a model of human fat metabolic disorders.

The authors of the second report, published in Nature Genetics, sought genes that, when inactivated, increased the C. elegans lifespan. Lee et al. followed up their RNAi screen with a classic forward genetics screen and, together, the results showed that worms with impaired mitochondrial and certain metabolic functions tend to be long-lived.

Apart from the important biological insights, Kamath et al. have given the community an important resource — the bacterial RNAi library can be used over and over again. Many more reports of screens such as those by Ashrafi et al. and Lee et al. are bound to follow, the results of which will provide a more complete picture of individual biological processes. The hope is that, as RNAi technology improves, similar systematic screens will also be feasible in mammalian cells.