With productivity goals soaring and investor confidence slipping, drugmakers are pressed not only to deliver improved new drugs but also to do so with remarkable efficiency. These new imperatives have catapulted robotics to the head of the pharma priority list.

"Pharmaceutical companies are spending less money on discovery R&D, so they need to spend it more efficiently," says Frost & Sullivan consultant Nate Cosper. "Microfluidics, robotics, and liquid-handling systems derive more value from what they already have."

Frost & Sullivan estimated that the world market for automated proteomics products alone would quadruple between 2002 and 2008, reaching more than US$750 million. The liquid-handling robot market, meanwhile, is set to double, reaching $1.216 billion by 2009, according to Frost & Sullivan. "These instruments release a huge bottleneck -- sample preparation," Cosper says.

Many companies are either entering this field or expanding their presence. Some companies have to automate to meet new needs. When researchers began applying DNA chips to large-scale compound profiling and toxicology studies, for example, Affymetrix started work on the GeneChip HighThroughputArray (HTA). Using available robotic components, this new system plugs the company's market-leading GeneChip technology into a standard 96-well microtitre plate format. By automating the slow steps, Affymetrix aims to lower the cost of large-scale studies while improving efficiency and accuracy.

Genetix Group is likewise keeping up with the 'more samples are better' trend. The recently launched QArraymax lets researchers quickly prepare large batches of their own glass, nylon, or other-material microarrays. The arrayer can hold up to 280 source microplates and can generate up to 450 slides or microtitre plates in as little as 10 minutes. "We've adapted our machines to spot on the bottom of microwell plates instead of glass slides," says chief scientist Julian Burke. Up to 100 protein or DNA spots fit into a single well in a 96-well plate.

Other companies are automating to expand established markets. Ciphergen Biosystems was one of the first groups to target protein biomarker development. But "when we first introduced the (SELDI) system, it was a chip and a reader," explains Ciphergen's Martin Verhoef. Customers had to create their own buffers and protocols. The process was faster than 2D gels, but that's not saying much. "It was a lot of pipetting," he sighs.

Since then, the company has focused on automation, using standard robots from Beckman Coulter and Eppendorf, and following the microtitre plate format. Now, "you put the centrifuged serum in the Eppendorf, and from there it is all automated," Verhoef says. Several hundred samples can be processed in a week.

Most importantly, Verhoef points out, automation "takes a lot of the operator out of the system." That means less error, and better results -- a key selling point.

The genomics boom spawned a wide range of new robots, as Agencourt Bioscience's laboratory illustrates. This reagent and service provider has two major advantages: its solid phase reversible immobilisation (SPRI) nucleic acid purification chemistry, and its focus on automation. "Every step is automated," says operations vice-president Brendan McKernan, "and it was conceived of that way."

The challenge at Agencourt is to accommodate a growing client base and continually expand the range of products. "It's easy enough to set up a pipeline process for one customer," McKernan says.

Agencourt says it can sequence an entire human genome in a year. "We'd like to do one a day," says Kevin McKernan, co-CSO and one of three brothers in management positions. The company's 40-plus Applied Biosystems sequencers run more than 20 hours a day. As one instrument nears the end of a plate, a digital sign flashes overhead, signalling that it's time for the next one.

Samples are prepped automatically with SPRI, on Beckman Coulter Biomek FX and Packard (now PerkinElmer) PlateTrak systems. "Every piece of automation is integrated into the LIMS," Brendan McKernan says. All the data feed into an Oracle database. Production, process management, and even proprietary reagent inventory supply are tracked electronically in real time.

Available tools are used whenever possible, but not at the cost of throughput. PCR preparation, for example, required some custom work. "You have to minimise every possible type of contamination," Kevin McKernan says. Otherwise, errant DNA may be amplified from dander and anything else that has floated into the hood or stuck to the pipette tips. When steel pipettes eroded under the strain of repeated sterilisation and use, Agencourt staff ordered custom ceramic-coated glass tips that could withstand the procedure.

Exelixis has also emphasised automation from the start, but its most unusual achievement has been the automation of functional genomics processes -- one of the major roadblocks in the path from new targets to drug. The company has done large-scale studies of mutations in several critical model species, including yeast, nematode, mouse, and zebrafish.

To create its genome-wide Drosophila melanogaster knockout collection, Exelixis had to generate, maintain, and propagate more than 30,000 strains. That requires transferring the fruit flies to new quarters occasionally. To move the flies, "you have to give an aggressive flip of the wrist," says Exelixis research head Greg Plowman. Daunted by the prospect of "flipping" about 30,000 tubes of flies once every month, the company designed a robotic system very similar to that used in high-throughput screening (HTS). "Instead of vials of compounds, the robots pick up 96 vials of flies, and put them into new containers," he says.

All of Exelixis' automated systems leverage standard instrumentation. "You don't want to build anything from scratch," Plowman explains. On the data side, "everything is Oracle- and web-based," he says. Robo-mania is helping Exelixis stock its pipeline with drugs against novel targets, at record speed.

Proteomics is another area where major technical breakthroughs are needed, as well as much more automation. Charles River Laboratories Proteomics Services (CRLPS) recently answered many of its dreams with the acquisition of an Advion BioSciences NanoMate chip-based nanoelectrospray system. "The nanospray is the most sensitive tool (for LC/MS-based protein identification)," says CRLPS president Jim Jersey. But the typical nanospray is not rugged and takes hours to run, according to Jersey. The Advion instrument saves time as well as sample, and provides researchers with a lot more information. "What took one to two hours takes five minutes," Jersey says. "And since you aren't time-limited, the mass spectrometer can be used to mine more content."

At the other end of the discovery process is another key milepost -- the process of lead compound selection. Screening is increasingly becoming ultra-high-throughput, allowing more than 100,000 compounds to be tested per day. Many big companies are spending heavily in this area: GlaxoSmithKline is reportedly investing about $200 million to establish several ultra-high-throughput screening facilities around the world.

"There has also been a lot of effort to automate parallel synthesis for combinatorial chemistry," says Wyeth's Jeff Paslay, vice-president of chemical and screening sciences. "But the other critical part is having ready access to those libraries."

Like most pharmaceutical companies, Wyeth is intent on building a better compound library. The company's goal is to have about 700,000 compounds in its screening library; it already has about 500,000. One major delay typically arises after the initial screen. "If you screen half a million compounds, and you have 1-percent activity, you need to go back and get the 5,000 compounds you want," Paslay explains. That task used to require weeks at Wyeth. Not only did compounds have to be retrieved by hand, they were also in powdered form, making it necessary to dissolve them before use. Besides a rapid retrieval mechanism, Paslay's group needed some type of liquid-sample storage facility.

That requirement raised a huge concern. "The compound library is one of the corporation's major assets," Paslay says. "Most such collections are worth tens of millions of dollars, at least. Maintaining integrity of the samples is essential." The compounds can't be stuck on any old set of shelves: Features such as temperature and humidity are critical, in addition to easy retrieval.

The solution to Wyeth's problem was a hybrid system developed in collaboration by TTP LabTech, a division of TTP Group, and TekCel. TTP contributed its Compound and Compiler modules. Samples are stored in microtubes in Compound. Compiler plunks those samples into 96-well racks, and delivers them to the TekBench, which plates them. TekCel's PlateServer then seals the plates before they are loaded onto the PlateStore mobile system, which can deliver them to remote sites.

In place since the end of 2003, the new system is an unqualified success. "It now takes less than half a day to get the compounds you want," Paslay says. "It's great. It's spectacular!" Most importantly, "it's really working," he says. That's critical to Wyeth, which aims to file two new drug applications for new molecular entities per year, starting in 2006. "The goal of every company is to try and increase the number of HTS campaigns," Paslay says. "Now we have time to do that."

Naturally, necessity gives rise to opportunities for automation. Syrrx developed its rapid-fire protein crystallisation system to gain perspective on important drug targets. The company concentrates on well-validated, yet poorly characterised, targets.

By addressing scale as well as speed, Syrrx has been able to reduce the amount of protein needed for crystallisation by a factor of 40. A 50nL drop of protein and equal amount of crystallisation solution are all that's needed to grow a crystal. "Once things are optimised, we can get crystals forming in a day," says Ken Goodwill, Syrrx's director of business development. To handle such tiny volumes with precision and speed, automation is essential.

The company has already run 8 million crystallisations, and expects to announce its first Phase I trial -- with a diabetes drug that targets dipeptidyl peptidase IV (DP4) -- later this year. The crystal-growing process is so successful that Syrrx has begun licensing the technology. In February, Fluidigm became its first licensee. "To our knowledge, this is the highest-throughput system in the world for this," Goodwill says.

Similarly, NextGen Sciences' ExpressionFactory emerged partly from the company's own need for purified proteins. "We need proteins to make biochips," says Grant Cameron, commercial director at NextGen. The company is exploring making protein chip-based diagnostics. "But there are many uses for proteins, including crystallography," he says. NextGen is now marketing the product as well as using it internally.

The platform automates everything from clone design to expression optimisation. One of the slowest steps in making proteins is inserting the gene into multiple vectors. ExpressionFactory uses Invitrogen's Gateway system along with other systems. The instrument makes it simpler to express many proteins simultaneously, and then check the results of all these experiments. "We are now five times more productive with the same number of people," Cameron says. One of the instrument's biggest selling points is the software, he says, which "lets you design and plan your experiments in silico before you commit to doing anything with the robot."

But automating complex laboratory processes isn't a cinch, and drugmakers are unwilling to compromise when it comes to quality control, flexibility, and convenience. That's why it's taken so long for many of these robots to make the grade.

For example, several years ago lab-on-a-chip technology was hailed as the next breakthrough set to transform R&D. Today, barely a handful of such products are available. "It's not as easy as you would hope to devise successful microfluidics," says Tony Owen, marketing manager at Agilent Technologies. "It's often easy to show feasibility, but quite a different matter to make a reliable and rugged product." Microfluidics is easily influenced by many factors, he explains, including contamination and surface effects.

One of the few lab-on-a-chip products is the BioAnalyzer, designed by Caliper Technologies and Agilent Technologies, which is so successful it has become the instrument of choice for many tasks, including preparing samples for microarray studies and PCR. Soon, Agilent will release a high-throughput bioanalyser, called the Automated Lab-on-a-Chip. "The BioAnalyzer is popular because it is fast, accurate, and it conserves sample," Owen says. The original instrument manages a dozen samples in one run; the Automated Lab-on-a-Chip will be able to do more than 1000, completely unattended.

Of course, automation can sometimes create another bottleneck -- downstream in data management. Genome researchers at the Broad Institute were some of the first to automate genomic processes, and have terabytes of experience with this issue.

Stacey Gabriel heads the institute's Human Haplotype Map Group, which aims to generate about 15 million genotypes a year using four Sequenom MassArrays and a recently acquired Illumina BeadLab System. The institute will generate that many genotypes again in disease studies. "One of the biggest things on my mind back four years ago, when we selected the first system, was automated allele calling," Gabriel says. Interpreting the data from such instruments has traditionally required countless person hours. "That becomes impossible with the amount of data we are generating," she says. "The MassArray nicely incorporated that feature." Besides automated sample handling and allele calling, the newer BeadLab System features an integrated LIMS and bar coding, both of which are a big help in data management.

Instrument makers are also starting to realize how important the software is that controls the instruments. Traditionally, instrument makers "have not paid enough attention to the ease of use of the software," says Josef Syfrig, Tecan Group content specialist in marketing communications. A major player in liquid-handling instrumentation, Tecan recently released new instrument control software called Freedom EVOware. The company says the new software will make it easier for anyone, from routine users to experts, to use the instruments. It also integrates easily with other vendors' tools -- an increasingly important consideration in this arena.

A startling number of slow zones have been speeded up thanks to automation. But technical bottlenecks remain. Wyeth's Paslay cites screening assay development. "You still need an assay that gives you the right answer, and you can't automate that," he says. And of course, robots also break down occasionally. "That's where the bump is," Gabriel says. Good service and software can address that problem, allowing researchers to quickly pick up where they left off, but some vendors are lacking in this regard.

In the end, robots can't actually find drugs. They are only as good as the scientists using them. But as long as they are giving researchers more time to think and plan, these automated systems are a welcome addition to the scene.