If you've recently knocked out a gene but your mutant has no obvious phenotype, then read on. Because a recent paper in Nature Biotechnology reports a new way to overcome the problem that many functional genomicists face — elucidating the function of a gene that produces no overt phenotype when inactivated.

To find phenotypes for such genes, Raamsdonk et al. delved into the Saccharomyces cerevisiae metabolome — its complement of ∼600 cellular low molecular weight, metabolic intermediates. Metabolites, like proteins, are functional properties of a cell and change according to a cell's physiological, developmental or pathological state. Because of this, the authors reasoned that the effects of some genetic mutations might be counteracted by metabolic fluxes that prevent overt phenotypes, such as altered growth rates, from occurring.

But there is a problem with this approach. How do you go from metabolic profile to gene function if there is often no direct relationship between the two? To overcome this obstacle, the authors developed a new approach called FANCY.

FANCY (for functional analysis by co-responses in yeast) is based on the measurement of the steady-state response of two variables to the change of a system parameter — that is, in this case, the change in concentrations of two metabolites in response to a mutation. The premise for this approach is that these responses should be similar in cells that have had the same functional response or pathway disrupted. In this way, matching the metabolic response to the loss of an unknown gene to that of a known gene should give clues as to the gene's function. The authors validated this approach by looking at two deletion mutants in the glycolysis metabolic pathway. The two mutants had no growth-rate phenotype but did have a metabolic phenotype when the concentrations of six metabolites were measured. Changes in the levels of these metabolites were similar in the glycolysis-deficient mutants, but were different to the changes in two respiratory-deficient mutants.

However, this approach is somewhat limited by the need to know which metabolites to measure. How could you predict this for genes of unknown function? So, to overcome this limitation, the authors went one step further and used nuclear magnetic resonance (NMR) spectroscopy to take 'metabolic snapshots' of cells. This approach can be used without knowing which metabolites a mutation affects. And Raamsdonk et al.'s stastistical analyses of the resulting NMR spectra could distinguish mutants with different metabolic phenotypes, and cause those with similar phenotypes to cluster together. These clusters can then be analysed by using known functional mutants to guide the analysis of co-clustering, poorly understood mutants.

This approach is now being used in a high-throughput screen of a library of ∼6000 S.cerevisiae mutants, each of which has a single open reading frame removed from its genome. Because this approach can reveal the function of non-metabolic genes — genes that are involved in protein biosynthesis, for example, the deletion of which would cause the cellular concentrations of certain amino acids to change — it promises to add to our growing knowledge of gene function, both in yeast and in higher organisms.