For four years now, Agilent Technologies (Sunnyvale, CA) has hosted an annual users’ conference on laboratory automation. The 4th Automation User Group Meeting was held at the Mark Hopkins Hotel in San Francisco, CA, from September 20 to 23, 2011. Key points of reports are: Automation of sample prep increases data quality and quantity while reducing costs, and laboratory productivity largely depends on design of experiments (DoE) and work flow. Today’s automated systems for chemical and biochemical assays, such as the Agilent BioCel, integrate modules and software from many vendors. Proposed next-generation systems should share a common industrial design for the modules to improve layout and servicing.
Laboratory automation: Focus on the biopharmaceutical industry
The meeting opened with an economic analysis of laboratory automation, with a particular focus on the biopharmaceutical industry, given by Dr. Peter Grandsard of Amgen (Thousand Oaks, CA). “Biopharm” is growing, especially since many new drug candidates are constructs of large and small molecules. Both regulators and developers demand a science-driven approach for development. He separated R&D efficiency from effectiveness, where efficiency measures the cost and time to introduction.
Big pharma, which includes Amgen, suffers from low productivity, as measured by launched products. Thus, instrument purchases are under cost-cutting pressure. The phrase "top of the line" is being replaced with “fit to purpose.” But their business is also global: Amgen wants the service level to be high and globally uniform.
Automation facilitates standardization of lab protocols across functions and along the R&D maturation path. For instance, as a small-molecule product candidate goes through research, preclinical, and development phases, the synthetic processes are scaled up to meet the needs of each phase. Scaleup means change. Grandsard pointed out that companies are working on unit synthetic processes that can be the same from the lab to the factory floor. The latter would just use many more of the units running in parallel. The example he gave was synthetic chemistry modules that can scale from mg/unit to kg/room by using the required number of units running in parallel. This approach might reduce concern that scaleup of products might change the product from that which was used in the original NDA.
Massively parallel instruments for research seem to work. The Broad Institute in Cambridge, MA seeks to develop and apply “systematic approaches in the biological sciences to dramatically accelerate the understanding and treatment of disease.” The evolution of genome sequencing at Broad was described by Ms. Sheila Fisher. Broad was one of the leaders in sequencing the human genome and follow-on genomes. In 2005, the Institute operated 117 ABI 3730 DNA sequencers (Applied Biosystems/Life Technologies Corp., Carlsbad, CA). Productivity was about 60 GB/year, which is about one genome, since high redundancy is needed.
In 2006, throughput was increased to 2.8 TB/year. By 2010, throughput increased still further, to 50 TB/year, and in August 2011, the rate was 500 TB/year and accelerating. This involves 51 systems (Agilent liquid handlers feeding the HiSeq 2000 [Illumina Inc., San Diego, CA]) that automatically analyze about 10,000 samples per month.1
For the most recent round of instrumentation, Agilent was selected as the vendor for the sample prep stage, which uses Agilent Sure Select Enrichment chemistry implemented on its Bravo automation platform. Ms. Fisher advised that full automation improves consistency and reduces contamination. “No touch” is the current criteria to reduce contamination. Acoustic liquid handlers from Labcyte (Sunnyvale, CA) are used to decrease plastic waste burden with air-driven pipetting.
Fisher pointed out that automation generally does not mean a reduction in full-time employees. However, when the automation is successful, one has the data and can then focus on improving the process even further, such as throughput, resolution, and data quality, or move on to distill knowledge from the data.
Scott Fulton of BioSystem Development (Madison, WI) lectured on the AssayMAP™ Microchromatography Platform for sample prep using automated liquid handlers such as the Bravo. Disposable plastic cartridges contain 5 μL of sorbent constrained by upper and lower support filters. As with solid-phase extraction, the particular sorbent can be selected from a variety of surface chemistries bonded to 15–100 μm particles. Using a special 96-well head, the Bravo precisely controls, aspirates, and dispenses with flow lower than 1 μL/min. Pmax is 20 bar. Fulton explained that a low liquid flow is essential in quantitative extraction. Mark Novotny of the J. Craig Venter Institute (San Diego, CA) pointed out that low flow rate dispensing nearly eliminated interwell contamination in his sample prep of DNA from single bacteria.
According to Hinnerk Boriss (Sovicell, Leipzig, Germany), estimating the unbound fraction of drug candidates is essential for predicting their pharmacodynamics. Sovicell developed the TRANSIL brain absorption assay for measuring the transport across a porcine lipid by layer. The assay correlates well with conventional dialysis methods, but is much faster, requiring less than 30 min on the Bravo liquid handler. Final quantitation is with RapidFire TOF-MS (also from Agilent) or with LC-MS, which is slower. Sovicell offers assays for liver binding as well.
Peptide samples can also be prepared with the Bravo liquid handler using PepTips from Glygen Corp. (Columbia, MD) in a 384-well format. Peptides (15–20 mers) are synthesized in microtips using FMOC chemistry. Run time is less than 24 hr. The peptides can be used in the tip or eluted into a 384-well plate for further use.
Integrated automation solutions and work flow organization
Laboratory automation instrumentation is a fragmented market. A few vendors, such as Beckman (Fullerton, CA), offer a limited number of workstations that focus on particular applications. Generally, the major vendors offer freestanding modules. Even a modestly complex work flow will usually involve components from two or more vendors.
Who integrates all of these components into a system? Some users might be adept at interconnecting the mechanical layout to comply with the work flow, but few would list software integration of instrumentation as a core competency. In practice, there are no small software problems. Writing drivers to interface from the system controller to the modules requires cooperation of the various vendors, plus technical competence in programming.
In prior years, Agilent and H-P had minimal involvement in systems integration or special instruments. For example, in GC, requests for specials were routed to the Channel Partner program. The Bio-Cel program from Agilent is designed to design, sell, assemble, and support custom-automated systems for chemistry and biochemistry. Agilent supports over 600 modules from nearly 100 vendors, providing mechanical and software integration via VWorks software.
The next generation of laboratory automation
Looking at the state-of-the-art, most of the workstations (such as the BioCel) seem to be a collection of components with a robot as the frenzied orchestra conductor. Each piece of apparatus looks like a kluge. Some units accept plates in the portrait mode, and others in landscape mode. Some have top entry and exit, with others on the side. Real estate around the robot is scarce and often difficult to access.
The industrial design for next-generation instruments should be looked at from an industry-wide perspective. Clearly there is room for improvement in current design: Modules could form a circle surrounding the robot, which would have an industry-standard inner diameter. Each module would receive and expel the sample plate in portrait format. The rear would provide access for loading and servicing. Heights could be quantized also, with 25, 50, or 75 cm to promote stackability. Widths could be 60o to accommodate six modules/layer. Robots should be extendable from near-floor to near-ceiling heights. Waste processing needs attention also.
Agilent is to be congratulated for organizing a balanced and informative program that included both chemistry and instrumentation. Agilent’s market focus is midmarket labs that run experiments of about 10,000 wells or smaller. This segment should grow rapidly, since many assays are being developed, which will increase applications. Also, funding should be more readily available for systems in the $500K to $5 million range. This is in contrast to ultrahigh-throughput screening systems that usually reside in core facilities serving the enterprise, often running experiments with a million wells.
1. http://investor.illumina.com/phoenix. z h t m l ? c = 1 2 1 1 2 7 & p = i r o l - newsArticle&ID=1435009&highlight; accessed Oct 7, 2011.
Robert L. Stevenson, Ph.D., is a Consultant and Editor of Separation Science for American Laboratory/Labcompare; e-mail: firstname.lastname@example.org.