After plant samples are ground into fine powder and sub-sampled to vials a number of different analyses are performed, including digestion studies, NMR analysis and GC-MS analysis. Automatic analysis systems exist and are common, however the preparation of plant tissues is still predominantly performed manually. Automating such processes presents many challenges but with large investment from European and US governments the incentives are high.
A thorough understanding of the process flow can be badly flawed if each highly repetitive process turns out to require a little additional time. If eight sets of 96 samples take a couple of extra seconds to cap or grind or settle the throughput can become horribly slow. Awkward as it may be experimental rigs or documented evidence should be used to provide a process speed analysis. Early attempts to define a process when full manual processing has not been confirmed requires a large degree of flexibility to be incorporated into the system designs. A small number of hypothetical protocols are then used to define acceptance testing. However, each implication and associated test method must always be defined, in case some or all of the flexibility is required at a later stage. It's all very well having a system where every variable can be altered in the protocol input file but understanding the problems re-familiarising oneself with it every time a new protocol is required must not be underestimated.
Grant, budget and tender timelines are never on the side of the project. Infuriating delays in the progress of these processes can be debilitating when the purchaser and contractor are finally given the go-ahead. The success of a system is highly dependant on the volume of quality robustness testing which cannot be rushed. Anything that can maintain momentum of significant but simple preliminary tasks that are undertaken during these delays can only help.
The sternest problems are found when attempting to automate the cryogenic grinding and dispensing of plant tissue. The first major problem is sourcing and integrating automation components that can operate at temperatures of -70°C. Often component datasheets state lowest tested temperatures which are based upon common applications. In practice a number of significant components required testing at -70°C to demonstrate performance. Sealing materials such as silicone, Viton and EPDM were tested but none proved effective at such low temperatures. Bearing components, lubricants, engineering plastics and electronics all required testing before a design finalised.
Accumulation of frost is also a show stopping issue as any moving components are quickly rendered inoperable due to frost. The solution is to seal the entire system and purge the wet air from the system, replacing the atmosphere with dried air; only when the relative humidity is below 5 per cent can the system be chilled. Sealing a bespoke robotic system requires careful consideration of cabling, glanding and operator access to ensure the solution is both sealed, ergonomically sound and suitable for maintenance.
The temperature gradients within the system cause thermal expansion and contraction, in-fact some components can move by as much as 2mm. The solution is very careful design to ensure that the expansion and contraction at every interface between components is managed. The thermal dimensional changes cannot be eliminated but they are measured and managed within the system design.
Picking, placing and manipulating consumables in a -70°C atmosphere is not trivial. Traditional pneumatic or electromechanical grippers do not function. Possible alternatives are electromagnetic picking or vacuum picking. In fact the most successful approach is to marry both technologies where possible. Picking with a vacuum is not preferred as carefully cooled dry is removed from the freezer, however as most consumables are entirely plastic often vacuum picking is the only option.
A problem common with any system which both grinds/mills plant material and weighs to 5dp (0.00001g) is vibration isolation. Grinding is performed by oscillating a vial filled with plant tissue and ball bearings. The frequency is around 30Hz and the amplitude as much as 30mm. This vibration is easily transmitted through rigid framework and severely affects the accurate analytical balance. As with the design of a system for thermal expansion and contraction, major interfaces between components must be considered as these give the designer the opportunity to isolate vibration. A 'vibration map' can be plotted from the cause to effect and a strategy implemented to minimise the impact.
Experience is the best tool for the implementation of new equipment. However, a thorough consideration of all the practicalities, human foibles and process risks is the next best approach as all these important issues are easily transferable from any other management tasks.
To summarise, it's the specifications that are paramount. These including working practices, purchasing, safety, scheduling, good practice and all the other influencing factors. The latest GAMP 5 (Good Automated Manufacturing Practice) manual gives a very thorough and up to date approach to every issue that needs considering. Even if the specifications you need and the system you want dose not warrant a full GxP approach it's probably because the application is tidy and controlled already. Typically the type of sample preparation needed for plant materials is complex. A really thorough understanding of each and every issue is required. However, with clear planning such a project is invaluable in being able to process enough samples.
Most suppliers of automation are good at leading a prospective purchaser through the learning curve to providing good specifications. It isn't a bad thing to start out with a wish list. This begins an iterative process whereby questions are asked and answers sought. Thre are always going to be dead ends which only serve to illustrate an impractical aspect. All too often a high throughput approach has to be abandoned as there simply isn't enough laboratory space for the numbers of samples or the sampling technique just isn't suitable for miniaturisation.success.
Enter 39 or at www.scientistlive.com/elab
Andrew Whitwell is with Labman Automation Ltd, Stokesley, North Yorkshire, UK. www.labman.co.uk.
