July 15, 2009 04:04
With the global economic turmoil going on and almost everyone seeing and many feeling the effects of the current economic climate, many scientists ask themselves how and where they will be working in the future. To put this into perspective I think it's worthwhile looking at what has happened to the auto industry that's currently in a major crisis in Europe and North America. Car manufacturing has seen two fundamental shifts in the last ca. 30 years: 1. Massive automation that came with the advent of sophisticated manufacturing robots and 2. Manufacturing disintegration and global outsourcing of part manufacturing and assembly. Both of these trends were driven by technology development and relentless pressure to bring down costs. The latter of course driven in part by us customers who want to spend less money for a car.
How is this related at all to protein crystallography? For one, in the past 10 years we have seen in our field the deployment of massive automation in many protein crystallography laboratories. Robots can set up protein crystallization trials and inspect them, requiring little human intervention. Once crystals are harvested they can be investigated at robotizised X-ray beamlines in ways that enable diffraction experimentation in fully automated ways. When user intervention is required, this can be done remotely, from any internet connected computer. And data analysis for structure determination can be done in automated ways as well. I have witnessed this myself in an impressive demo that Wladek Minor gave earlier this year at the "Enabling Technologies" workshop at the NIH. Like a monkey Wladek had me klick two buttons on his laptop running HKL, only to build a complete model, starting from diffraction images, within the duration of the presentation that he gave. All of this automated technology has increased throughput and structure output, with ever diminished need for human intervention. This is welcome news for us tax-payers and consumers of modern medicines who have an interest in spending our money most effectively and paying less for pills. So it is good that "the cost per structure has fallen continuously; the range of previously unknown folds and protein families has been vastly enlarged" as stated by David Eisenberg in the NIH report.
X-ray protein crystallography is arguably a very technical, and in high-throughput mode a non-hypothesis driven exercise, especially when played in a high-throughput structural genomics environment. While the result of our work, protein structure models, are vehicles to generate new hypotheses, the process can be executed for many protein targets with this automated machinery, requiring no creativity and little human input.
And what about the theme of disintegration and outsourcing in our field? X-ray crystallography has become the ‘killer-application' for lead compound optimization in early stage drug discovery and had created a substantial demand of highly skilled crystallographers. Big pharma is in the midst of a deconstruction phase and many players are re-organizing their discovery efforts. Leaving complete discovery efforts to the biotech industry has become a standard play. Presently the financial markets though are not in favor of funding risky new biotech startups. But even a more positive financial climate may not nourish the biotech sector since it has become clear in recent years that biotech as a whole has yet failed to deliver on its promise (also here). The drug discovery programs remaining in Pharma are poised to rely on outsourcing entire discovery projects or project modules to contract research organizations. And as long as the cost to pay a scientist in Asia is a lot less than that in Europe or North America, it makes economic sense to tap into these resources. After all, in today's flat world it hardly matters where the infrastructure is set up and ‘tending the machines' is a transferrable skill. Crucially, it will not be possible to tell the difference between protein structures determined at the ESRF or the BSRF. There's nothing wrong with these changes per se, I'm wondering though what the crystallographers in Europe and North America can expect to be doing in the future.
Of course you could say that there are those difficult targets such as multi protein complexes and membrane proteins that are not (yet) amenable to automated structure determination. I'd argue though, that the difficulties in handling such targets are left to molecular biologists and protein biochemists that are involved in the ‘sample preparation' aspect of structure determination. Once there is well-behaved protein sample and crystals the machines can take over again.
In the grand scheme of things we're witnessing X-ray crystallography maturing as an industrial technology and this typically creates new opportunities. While I don't see yet how these new opportunities manifest themselves, I'm wondering what's left to do now that the machines and cheaper labor are taking over. My gut feeling is that many crystallographers will need to do some soul searching and make up their mind if they want to (i) ‘work the machines', (ii) teach the next generation of crystallographers, (iii) further develop and implement the technological aspects of crystallography or, (iv) start utilizing the information generated and address real biological questions. And there will always be a place for the expert that is really good at solving the rare and very tough crystallography problems.
Plenty of work to do it seems. This brings me back to the title: who will be needing us crystallographers in Europe and North America? You tell me.