Without new synchrotrons, US structural genomics may depend on automating existing facilities to meet its goals. But without an influx of interested physicists and engineers — and the funding to support them — speeding up the process of discerning a protein's three-dimensional structure may prove difficult.

The automated gene sequencer revolutionized genomics. But no single machine, by itself, will greatly speed up structural biology. Working out a protein's structure is a more complex process, with many more stages, says Peter Kuhn, assistant professor at the Stanford Synchrotron Radiation Laboratory.

The protein must be purified, crystallized, mounted in the beamline, probed by the light, then reconstructed by computer. Potential bottlenecks occur at every stage — and each stage may need different combinations of expertise to speed it up. For example, automating sample mounting in the beamline requires engineering skills, whereas finding and centring the sample within the beam needs computational skills.

Different US groups are working on separate stages of this problem. For example, Stanford is in the nascent stages of automating sample mounting. Once individual steps have been automated, systems engineers will need to evaluate how the entire process fits together, Kuhn says.