Automation and robotics are increasingly freeing life scientists from the need to baby-sit their experiments. The result:
increased productivity and lower costs for laboratories and more creative time for individual researchers.
|by Peter Gwynne and Gary Heebner
|•||Shorter Time, Lower Cost|
|•||How Much Automation?|
|•||Varieties of Automation|
|•||Flexibility Vs. Specificity|
|•||Decreases in Volume|
|•||Levels of Sophistication|
|•||Lab Management Systems|
|•||From Drug Discovery to Clinical Use|
|•||The Next Steps|
Progress in understanding the nature of life stems in large measure from the discovery and development of fundamental tools and techniques, such as enzymes that splice genes, the polymerase chain reaction (PCR), and means of sequencing the genomes of various organisms. Less glamorous, but equally important, are developments such as high throughput sequencing and microarrays that speed up the work carried out in the laboratory. In recent years, life scientists have started to benefit from another mechanistic advance: the introduction of laboratory automation and robotics to the R&D lab. Researchers are making steady progress advancing their science with new tools, from kits and reagents to instruments and systems, that can prepare samples, run experiments, and analyze results.
Automation of routine laboratory procedures, by the use of dedicated work stations and software to program instruments, already exists for basic laboratory routines. Some laboratories have even enlisted the services of fully functional robotic systems to replace jobs once done manually during an eight-hour shift into procedures that require very little human intervention and operate (at least in theory) 24 hours a day, 7 days a week, 365 days a year. "Any simple, repetitive task — such as pipetting, moving plates around, and various types of assay — can be automated," says Al Outhouse, senior applications scientist at CRS Robotics. "We see this as all part of the industrialization of biology," adds Clifford Baron, director of marketing for global services and solutions at Applied Biosystems.
Laboratory automation and the growing emergence of robotics have transformed the typical workday for many individual scientists. Thanks to the creativity, imagination, and hard work of researchers and companies in this field, scientists can set up, run, and analyze the results of experiments in a fraction of the time they needed in the past. They can also accomplish the tasks with less hands-on intervention than ever before. As a result, associate scientists and technicians who used to spend their days performing tasks of tedious repetition now have the time to think creatively about the implications of their experimentation and to design effective follow-up projects or develop alternative approaches to their work.
At the corporate level, and particularly for firms involved in drug discovery and clinical diagnostics, automation and robotics have significantly increased productivity and lowered costs. For executives of those firms, squeezing the maximum efficiency out of every department, including the R&D lab, is a constant issue.
Beyond cost-saving, two main factors have encouraged the growth of automation and robotics in life science laboratories. "What's driving acceptance of robotics is the fact that the cost of error is very high, in a scientific paper or developing a drug," says Baron. "Even fairly low error rates can have a profound impact on the conclusions you make downstream based on your data." Mike Olive, director of molecular biology at LI-COR Biosciences, develops that thought. "The average researcher in the lab needs consistent quality," he explains. "You assume that by taking out the human element you will get more consistency." In addition, organizations increasingly expect their research scientists to concentrate on their areas of expertise. "Life science laboratories in the industry and in basic research want to focus on their core competencies," says Gaby Bachofner of Swiss firm Tecan Group Ltd. "Therefore they increasingly expect complete solutions from their automation partners."
Applications for lab automation range from the use of multitip pipetters to fully automated robotic stations for a high throughput operation. The amount of automation that any lab requires depends on its situation. While an academic research lab may choose to use only some instruments to increase productivity and eliminate a tedious task, a drug discovery unit in a pharmaceutical company will probably want to automate all phases of its research.
Scientists can automate many basic laboratory procedures with minimal effort. Dispensing cell culture media into flasks, filling multiwell plates for assays, washing, rinsing, and applying reagents in an immunoassay all present opportunities for automation to one degree or another, depending on the number of samples being processed. German company Eppendorf AG, The Hamilton Company, and VWR International are among the many companies that offer user friendly instruments for such tasks.
Laboratories that work with DNA sequencing and genomics have a more intensive need for automation. Sequencing DNA fragments can involve a large number of repetitive steps on huge numbers of samples. So several suppliers of DNA sequencing and analysis instruments have either developed automation capabilities in house or have linked up with companies experienced in automation to design instruments that can run more samples with less human intervention. "We have a robotic pipetting station that interfaces with a number of other instruments," says LI-COR's Olive. "Our SagaGT software is aimed at automating the process of microsatellite analysis and provides automated data analysis and allele calling. People like it because it frees them up to do other tasks while the software is working on the data."
Any laboratory manager who wants to automate operations must decide on which semi-automated or fully automated system to purchase. The manager should base that decision on several basic factors: why the lab requires automation; what assay format the laboratory will use; what level of technical support it will need; and what potential disadvantages might surface after installing such a system. Having decided to automate, the manager must then examine what's available on the market.
Manufacturers of laboratory automation and robotic equipment emphasize that their products don't come in a one-size-fits-all format. Rather, they design specific devices for the specific needs of specific laboratories or researchers. "We look at it from two aspects," says Tuula Jernstrom, marketing manager for Finnish company Thermo Labsystems Oy. "People want higher throughput to screen more samples or handle more plates. Or they want to automate the manual part of their work."
John Comley, product manager for liquid handling of high throughput screening at PerkinElmer Life Sciences, makes a similar point. "You have to differentiate between full automation and partial automation," he says. "Within the pharmaceutical industry, high throughput sequencing laboratories have pioneered laboratory automation. They talk about 'drug discovery factories' that are entirely automated. But that idea is not universally accepted. Other labs have more of an idea of 'personal robotics.' That term denotes a smaller work station type of environment."
Not every life science lab needs a complete automation system. But even a little can prove useful. "The average academic researchers may not need as much automation. They have graduate students and postdocs, and they don't do the kind of throughput that needs automation," says LI-COR's Olive. "But everything that makes life easier will help."
Scientists are applying automation to a growing number of tasks. "Applying automation to primary screening began aggressively about seven years ago," recalls Chris Neary, manager of automated solutions, strategic marketing at Beckman Coulter, Inc. "Currently, owing to the increased number of hits generated in primary screening labs, there has been more focus on and greater acceptance of applying automation downstream to secondary screening assays."
Similarly, automated systems are expanding from genomics to proteomics. "High throughput mass spectrometers cannot be fed with sufficient samples of the required quality," says Tecan's Bachofner. "New solutions to improve reproducibility, increase throughput, and boost sensitivity for the separation and purification of proteomes are badly needed. Over the past years pharma research has invested heavily in high throughput screening [HTS]. That has created new bottlenecks downstream in cell based assays. New work stations to run these complex assays at high throughput and affordable cost are now hitting the market."
Automation has also reached beyond the research laboratory. "In clinical trials we're seeing a range of applications for bioinformatic packages associated with automation," says Ken Kirsten, senior marketing manager with SPSS Science.
To deal with the changing demand, vendors of lab automation equipment offer a range of products and services. "The days of one-trick only automation in research labs are limited," says Sven Bülow of Eppendorf. "The challenge is configuring components to allow maximum flexibility. Modular systems will allow scientists to tackle several different lab processes in one automation platform."
Thus manufacturers aim to give several of their lab automation products the maximum flexibility in application. "We're working to broaden the range of assays we can do on our systems," says Dan Roark, vice president of liquid handling and robotics for Packard BioScience Company, recently acquired by PerkinElmer, Inc. "Machines are now becoming more adaptable to doing different things," agrees Dave Hansen, product manager, instrument products for The Hamilton Company. Makers of robotic systems have the same experience. "Robotics is considered a general tool," says Hansjorg Haas, senior vice president of sales for global operations at CRS Robotics. "But people start to validate it for particular tasks and assays."
Illustrating the diversity of the market, however, some large customers now want systems that can carry out a single set of experiments consistently. "We're seeing growing demand for systems that can run an ELISA assay every day, seven days a week," says Rob Donoho, Beckman Coulter's manager of strategic marketing.
Technical advances in somewhat unexpected areas are helping vendors of lab automation and robotics systems to anticipate and meet their customers' needs. "One of the most significant changes we've seen is that improvements in the molding of plastics are getting to the point at which we can create plastics to handle smaller liquid volumes. That makes it possible to manufacture perfectly formed, disposable pipette tips and 384- and 1,536-well microtiter plates," says Donoho. "Assays used to be pushed into automation by customers who needed to go faster," adds his colleague Neary. "Now the plastics makers and other vendors are beginning their developments with the expectation that their products will be targeted at the automation customer. They see the market trends and they know that the customers will want automation; so they're proactively doing it and producing better, automation-ready products."
Vendors of robotic systems face one significant issue. "It's a matter of standards," says Haas of CRS Robotics. "Look at two-dimensional gel electrophoresis. That was difficult to achieve because of the lack of standards." Outhouse, Haas's colleague, argues that the issue lies with the customer. "Labs that want to robotize tasks will set the standards," he says.
Scott Vander Woude, director of marketing for the product group at BioRobotics, Ltd., a British member of Apogent Discoveries that develops robotic systems for work with microarrays, emphasizes both the need for standards and the difficulty of setting them. "We're doing wet chemistry protocols and training classes in Cambridge, England," he says. "You put those people in the low humidity and cold climate of Cambridge, Massachusetts, and everything changes. So you need to set standards that will work regardless of external conditions. When they become available, microarray fabrication will reach its fullest achievement level."
A related issue is the need for compatibility among laboratory robots. "There's a need for purchasers to make sure that their robots are compatible with everything else," says LI-COR's Olive. "In fact integrated systems are coming along. A company making an ELISA robot may not make other robots. But it will want to ensure compatibility with other companies' products."
A significant trend that influences the design of lab automation systems is the decrease in the volumes of samples to be squeezed into multiwell plates or other devices. Researchers have learned how to work with microliter and even nanoliter volumes. By reducing the amounts of reagent required, that approach lowers the cost per test and the expenses related to waste disposal. But designing pipetters and work stations that can handle such small volumes poses some significant challenges.
In particular, automatic pipetters for microliter volumes must ensure that each device delivers the entire sample volume. The retention of even a minuscule amount of sample in the pipetter tip — an all too common occurrence — can create significant errors. "Customers are looking for reproducibility of small volumes even when using samples with widely differing characteristics such as different viscosities or temperatures," says Eppendorf's Bülow. "In a research setting, this should ideally be achieved without the need for recalibration. The patented technology of our Nanozyme system addresses these issues in the nanoliter scale for the first time."
At the one end of the liquid handling spectrum, hand-held pipetters have advanced from a rubber bulb attached to a glass pipette to sophisticated electronic instruments that use disposable tips and deliver the tiniest volumes imaginable with relative ease. Some of these devices can be programmed to deliver exact volumes of liquid on a repetitive basis, which can be ideal for preparing a group of tubes in a rack or wells in a microwell plate. In addition to Eppendorf, Drummond Scientific, Jencons Scientific, and Rainin Instrument Company offer these devices, which include 4-, 8-, and 12-tip multichannel pipetters. Tango Liquid Handling Systems from Robbins Scientific offer quick change dispensing heads and low-volume pipetting for 96 or 384 channels at a time.
Hamilton Company has developed technology to handle problems inherent in the use of disposable tips. "A groove inside the tip has an O ring," explains Hansen. "The mechanism releases itself to drop the tip off without spraying aerosols around. And our MicroLab Star system can spread the tips unevenly, so that you can pick up an uneven distribution of samples. That's common in cherry-picking for applications in drug discovery. In addition the unique positive attachment mechanism for the replaceable pipetting tips allows detection of small pressure changes inside the tips for both pipetting precision control and detection of abnormal situations such as empty tubes of solids blocking the tip."
Filling, washing, and rinsing many tubes or multiwell plates can also be a tedious job, although essential for most laboratory work. Laboratory managers can now purchase small systems for the preparation work once done by hand. These instruments can wash and rinse several wells of a multiwell plate at once, helping to reduce the tedium of such a task. They are ideal for preparing a small number of multiwell plates.
Larger numbers of plates require more sophisticated instruments. Fully and semiautomated work stations can prepare or manipulate hundreds of plates in a single day for high throughput work. Work stations are usually dedicated to perform a fixed set of tasks, such as washing, rinsing, or evacuating solutions from a sample well. These stations often require the operator to refill them with empty plates when needed to remain in active operation.
Robotic systems are even more sophisticated, as they can perform many of the tasks that would normally be left for a human being. These tasks might involve gripping a tube or other object and emptying its contents or moving the object from one environment to another. Robotic systems of this type, produced by Zymark among other firms, are both smart and flexible. They can be set up to perform one operation and easily modified later to perform other tasks.
CRS Robotics has just introduced a robotic system that works with microwell plates at least five times faster than other devices. "We've taken a much more parallel approach to the processing of the plates," explains applications scientist Simon Foley. "Each individual instrument is addressed by its own small mover. They're all linked together. The system is significantly faster. It also allows you to introduce additional instruments when you want to increase throughput." Adds Haas: "You can always reconfigure the system when your application changes."
Tecan, meanwhile, aims to identify bottlenecks in liquid handling and to apply its core competence in robotics to the problems. "We deliver a complete solution with dedicated protocols and, in certain cases, with disposables and reagents," says Bachofner. "With this business model we were able to gain a strong presence in genomics for high throughput DNA preparation. We have also founded a subsidiary dedicated to proteomics with the recent launch of a product that addresses the major bottleneck of sensitivity."
Another area of application for automation and robots involves microarrays. A microarray is a regular pattern of spots or features placed or spotted onto the surface of a slide made of glass or other inert material using special robotic systems called microarrayers. These chips can hold up to hundreds of thousands of features on a single slide. Microarrays can be used with DNA for gene expression studies or with proteins for proteomics work and drug discovery. Alternatively, they can hold several reagents for a battery of diagnostic assays.
Microarrays can be made in a laboratory or purchased ready to use. Instruments can lay down small samples onto the surface of a slide with great precision. That allows the spots or features to be located in very specific positions on the slide. The microarray is then exposed to a sample or reagent. The slide can be processed after exposure to a sample and read in a special instrument or scanner.
"Microarraying is very similar to PCR," says Vander Woude of BioRobotics, Ltd. "PCR was the black magic voodoo at one time. Now you can pick it up out of a box and buy all the components. Microarray tools for researchers are not fully developed and available yet. Scientists don't commonly know where to get them. At BioRobotics we have kept our focus on providing the instruments for array fabrication. We have designed an all-inclusive system that controls the environment and meets the needs of scientists. Our goal is to be as flexible and helpful as we can in the process." Other companies that specialize in the fabrication of microarrays include Amersham Biosciences and MiraiBio (Hitachi Genetic Systems).
A cousin to the microarray, lab-on-a-chip technology, has received a great deal of attention in the past several years. Companies that have invested heavily in this technology, among them AVIVA Biosciences, Cepheid, and Gyros AB, have taken various approaches to miniaturization with chips and sometimes other devices such as compact discs. Cepheid, for example, is developing fluidic cartridges to perform the complex steps of DNA extraction from a variety of samples automatically. Based on leading edge microfluidics technologies, the company's disposable, single-use cartridges are designed to perform functions that range from sample containment and delivery to integrating the steps associated with DNA extraction, amplification, and detection in one system and one procedure, all starting with a real-world specimen.
Assaying also stands to benefit from automation. Most assays require a purified cell extract, which can take several hours to prepare and must be done with great care to avoid altering the intracellular contents of a dynamic and living cell. In addition to degrading or changing molecules with the mechanical forces that might be used to break open a cell, one also must be careful not to allow enzymatic degradation of proteins and nucleic acids via native DNAse, RNAse, and protease molecules.
Several companies have responded to this concern by creating systems that allow intact living cells to be examined in cell based assays. BD Biosciences, Cellomics, and others have designed systems that can process large numbers of living cells under relatively natural conditions to examine molecular interactions within cells. These systems expose cells to a compound of interest to determine whether any interaction occurs with the living cells. Fluorescent tags often allow the interactions to be detected.
Packard BioScience has developed a cell based assay system for high throughput screening that it calls the ImageTrak. It represents an application based on the company's PlateTrak platform, an automated microplate processing system that includes liquid and plate handling and works with batches of 50 to 400 plates. "PlateTrak is a modular system," explains Roark. "Every 10 inches you can have another module. We can have up to 16 process modules, such as washers, pipettes, and filtration devices." The ImageTrak system is based on a fiber optic contact imager. "We can use it to focus research efforts for GPCR receptor screening, including calcium and ion channels and membrane potential assays," Roark continues. "The system as configured is all inclusive. All the plate handling is tied to our system."
Beckman Coulter has developed its own approach to improving the productivity of automating assays. "Automated assay optimization has been widely accepted by large pharma companies. Therapeutic groups are not allowed to put an assay on an HTS system until they have optimized the assay using statistically designed experiments," says Neary. "It saves both time and money. It speeds up the assay development process by employing fractional factorial design and then automatically programs all the pipetting functions. What used to take three weeks to go through the design of experiments can now be done in an afternoon using this solution. It literally saves thousands of dollars in time and resources while greatly improving the performance of the assay."
The increased use of high throughput systems means an extraordinary increase in the amount of data generated in life science research laboratories. In those circumstances, says Oliver Bell, product line manager, sample preparation systems at Applied Biosystems, "the tracking of data associated with a particular sample is important. You have to track each sample on the instrument itself and also the location of data associated with the sample. We work on data management through the entire process."
That type of tracking often requires laboratory information management systems (LIMS). A typical automated system can log in samples, create batches and work lists, and group samples by test or other common attribute. A LIMS can generate schedules for testing and tracking samples in a clinical lab as they move from one department to the next. It can also log data from testing instruments, thereby increasing productivity and helping to eliminate errors. Accelerated Technology Laboratories and LabVantage Solutions, among other companies, offer LIMS. These systems provide good examples of using automation in an integrated fashion, communicating with many instruments in a laboratory or even in several laboratories in a facility. "The system can be very simple, filling in plates only. Or it can include several instruments, such as readers, washers, dispensers, and incubators," says Jernstrom of Thermo Labsystems.
Automation permits researchers to use large numbers of samples. With such numbers comes a mountain of data that can be overwhelming for individual researchers to decipher. "I think that data management is actually the gating factor in determining success," adds Dave Levy, vice president of product management at NuGenesis Technologies Corporation. "The huge volumes of data mean nothing unless you can draw conclusions and knowledge from it." NuGenesis recently announced the 5.1 version of its patented NuGenesis® Scientific Data Management System (SDMS) suite of products to provide as much information as possible about scientists' data. "We added the ability to launch applications from our system and bring in related data from other systems," says Levy. "Through NuGenesis SDMS, scientists can view and use their data regardless of proprietary file type for a streamlined collaboration and decision process."
Bioinformatic programs are designed to gather, store, and analyze large volumes of data. Suppliers such as DoubleTwist, Entigen, Accelrys, Spotfire, and SPSS Science provide several software packages designed to simplify a scientist's work in handling data. Entigen, for example, offers the BioNavigator program. Its graphical user interface is especially straightforward, providing fast access to analysis programs and data and accommodating users' work-flow preferences.
"These packages are absolutely critical at this stage of the game," says Kirsten of SPSS. "You're looking for patterns among many, many variables. You're trying to compare genes and experiments at the same time. This is classic data mining."
The basic approach to data mining clusters samples into groups that have some common characteristics. SPSS takes the process a stage further. "Our packages not only look for patterns in the data but also start to predict trends and events," says Kirsten. Our Clementine package does clustering, logistical regression, and prediction as well as machine learning. You'll be able to make predictions with the new data. It makes data mining feasible for people without a background in bioinformatics."
The pharmaceutical and biotechnology industries have driven automation as well as benefited from it. Drug discovery units in particular have taken the lead in automating and robotizing their research laboratories. Processes that used to take several months can now be carried out in a single day. Most of this gain in productivity stems from high throughput screening or ultra-high throughput screening instruments that eliminate many manual operations, reduce human error, and make it feasible to work with minute volumes of sample and reagent. This is especially important for samples that are very valuable or in short supply. Here, a mistake that causes the loss of a sample or experimental result may elicit an extremely high cost.
As the nature of screening work for drug discovery changes so does the related automation. "Most of our HTS work in the past involved detection technology," recalls Comley of PerkinElmer Life Science. "Now our focus, as a total solution provider, is on making work stations that fill a slightly different niche. The main emphasis here is making the link between reader and liquid handler modules seamless so that the components are truly 'plug and play,' giving users maximum flexibility and the ability to rapidly configure multiple applications-oriented work stations."
Drug discovery teams aren't the only groups in life science to benefit from automation. "Automation has been a large part of the instruments in the clinical diagnostics laboratory for years and years," says Robert Stoy, vice president of laboratory systems architecture for Beckman Coulter. Abbott Laboratories, Baxter Healthcare, Roche Diagnostics, and others have made great strides in designing instruments and reagents ready for use in clinical laboratories and hospitals where consistency, ease of use, and time to diagnosis can mean the difference between life and death for any patient.
An annual international event on high throughput screening (HTS) and miniaturization technologies organized by IBC USA Conferences Inc. will take place from March 4 to 6 in San Diego, California. "The event details all the critical elements of HTS and miniaturized screening in a new, three-day format," says Ellen King, IBC USA's project manager.
"The main conference includes a plenary session with application-oriented talks on 'Assay Development, Miniaturization and Validation' and sessions on the nuts and bolts of screening and of compound logistics and quality," says King. Preconference symposia include Screening Informatics and Data Mining, Advances in Cell Based Assays and Cell Based Screening, and Liquid Handling Challenges and Solutions in an HTS Environment. Several new technology workshops will also be offered.
A site tour to Discovery Partners International will provide an inside view of the company's laboratories and microarrayed compound screening systems.
For further information you can check the event's website, www.lifesciencesinfo.com/screentech. Alternatively you can phone 508-616-5550, fax 508-616-5533, write to IBC USA Conferences Inc. at One Research Drive, Suite 400A, Westborough, MA 01581-5195 or e-mail .
The trend has quickened recently. "What has gained some steam in the past three to four years is automation of the part of the work that involves moving blood and serum around the lab from one instrument to another," Stoy says. Why? "Transporting tubes of blood needs a lot of people," Stoy continues. "If you can get the robots to do it, you release people to do more important things. We're looking at robotic instrumentation to help process the blood tubes with the existing instruments we have in the lab."
Jan 'strup, manager of clinical screening at PerkinElmer Life Sciences, takes a similarly upbeat view. "On the clinical side we're seeing a combination of platforms using linear robots," he says. "We're working on two platforms. We have the idea of putting the sample in and leaving it to work."
How will laboratory automation change in the next few years? Will tomorrow's drug discovery lab be populated with mobile intelligent robots to the exclusion of human scientists? Rodney Brooks, director of the Artificial Intelligence Laboratory at MIT, thinks not. "We design mobile robots for unstructured environments," he notes. "Laboratories are very structured." Nevertheless, Brooks sees a future for artificial intelligence in the research lab. "Rather than filling up with robots, labs will go inside the box," he predicts. "Scientists will do experiments on a chip, using microelectromechanical system [MEMS] techniques that will lead to vastly more complicated machines able do a lot of the different processes currently done by bigger machines. Algorithms being developed now will help robotic-type decisions to be made within the chip. There is intelligence, there are perceptual algorithms, and there is MEMS. So we will have microrobots — not Hollywood type robots — in the lab."
Laboratory automation has already made dramatic changes to the way in which researchers approach their work. It will clearly provide new tools for increasing productivity in the future. Advances to come include programmable, automated work stations that can perform a multitude of tasks, extremely sophisticated robotics that can perform tasks once restricted to humans, and artificial intelligence systems on chips that can learn from their experiences much like ordinary humans. What is ultimately possible seems to be limited only by the creative genius of researchers working in the laboratory to define their needs and companies focusing on providing creative and user-friendly solutions to those problems and challenges.
Peter Gwynne is a freelance science writer based on Cape Cod, Massachusetts, U.S.A. Gary Heebner is a marketing consultant serving the scientific industry, based in Foristell, Missouri, U.S.A.
This article was published
as a special advertising supplement
in the 18 January issue of Science