New advancements in scientific software allow researchers to create virtual workcells with lab automation robots.By Alan H. Katz, PhD, Chief Scientific Officer, Hudson Control GroupRecent software advancements have made it easier to automate almost any procedure carried out in a research laboratory. It is now possible to design "virtual workcells" from a small, dedicated collection of liquid and microplate handling robots to perform the simplest or most complicated procedures. This article shows how the virtual-workcell approach makes it possible to build a flexible system for testing drug candidates with a wide variety of protocols. In addition, various molecular biology procedures often used in the manipulation of proteins and nucleic acids can similarly be run in a single workcell. Now that the setup of new protocols is so easy, automation can be applied to many tedious processes that previously were not considered possible.
Instrument setupIn choosing and setting up the equipment, maximize flexibility by including all of the equipment that one might need, and making sure a microplate can be readily transported between any two pieces of equipment. Design the layout so that each instrument is easily accessible for one-off, manual experiments. Two components that allow plate movement among instruments are robotic arms (e.g., PlateCrane from
Hudson Control Group), and moving tracks (e.g., Hudson's LabLinx). Instruments that are frequently used in sequence benefit from contiguous placement along a track. Other components, and those that cannot be adapted to tracks, must be within reach of the robotic arm.
Biological assay screening systemThe VaryScreen system (Figure 1) was designed to carry out a large number of screens normally encountered in drug discovery research. A multimode reader (e.g.,
Biotek's Synergy 4 or
BMG LabTech's PheraStar) supports the vast number of protocols involving fluorescence intensity, polarization, and resonance energy transfer (FRET), as well as AlphaScreen (
PerkinElmer), ELISA, luminescence and UV/VIS absorption-based assays.
The system includes a multi-channel dispenser (MicroFlex), an automatic pipettor (SOLO), and a searchable stacking system connected by a LabLinx track. In addition to the multimode reader, the system also contains an automatic incubator (Liconic STX40), a microplate shaker nest, and a microplate washer (ELx405) accessible to the PlateCrane robot arm.
Proteomics and genomics workcellA larger ProLink system (not shown) can be used for a wide range of molecular biology functions such as protein expression and DNA purification. The system contains a colony picker, plate sealer, centrifuge, thermocycler, and an automatic vacuum filtration system in addition to the components in the VaryScreen System above.
Software integration and virtual workcell designHudson Control Group's workcell scheduler software, SoftLinx For Systems, is also a tool for building virtual workcells from an integrated workcell, such as the systems described above. Software interfaces of all the equipment are first loaded into SoftLinx. Some of the equipment will require individual configuration, such as pinning down the location of microplate nests, or the type of plates and pipette tips being used. After this is complete, SoftLinx is sufficiently aware of the system to make it ready for protocol design.
The user designs virtual workcells by simply by building process flowcharts from the SoftLinks' Method Editor. The Editor includes icons for each instrument, which are dragged into the growing process, and appropriate parameters are defined (such as dispense volume or incubation temperature). In the simplest cases, the entire protocol can be included in a single process, such as the bicinchoninic acid (BCA) UV protein assay (Figure 2). More commonly, a protocol consists of a series of parallel processes, each focusing on a particular portion of the method.
Each process can be set to start as soon as the protocol is run; or it can be set to wait until any number of conditions are met, such as the presence of a plate in a particular instrument nest, the identification of an active compound (Example One), after a number of identified hits is reached (Example Two, page 36), or after another process is complete.
Processing logic and smarter automationMany frequently used protocols can be automated with the set of tools just described. However, the user can introduce programming logic to greatly increase the power and flexibility of automated protocols. For example, at the start of a process, one can create and set global variables so that every process can be aware of the current state of the entire system. A Boolean variable can remain false until a condition is met, or a desired result is obtained. A numeric variable can be used to keep track of how many times a particular plate has been washed, or the number of active compounds identified (Example Two). To take full advantage of these variables, an "If-Then-Else-" conditional tool is included to allow a user to control how a protocol proceeds based on various conditions. A "While" statement, another basic programming construct, allows portions of a process to continue (or stall) as long as a logical statement is true.
To incorporate flexibility and levels of intelligence into a protocol, one can include user scripts. With full access to Microsoft's Visual Basic for Applications (VBA), one can write very detailed programs that can carry out all sorts of numerical, string, or even inter-application communication. These scripts allow an automated process to analyze the results being obtained on-the-fly, and react accordingly. A few examples are included below to demonstrate the sort of functionality that can be added to a protocol.
Example one: Re-reading a microplate only when hits are foundOne can immediately perform additional steps on samples that show interesting results in an initial screen. These steps can vary in complexity from a simple re-reading of the plate, to a submission through an entirely different assay protocol. Regardless of the nature of this follow-up test, we first need to establish the presence of a hit in a plate by combining a user script with an "If-Then-Else-" control statement. In the simple example (Figure 3, page 34), a spectroscopic study has been carried out on a BioTek Synergy 4 plate reader. After a run, the Synergy 4 creates a text file containing the results from a procedure defined with BioTek's GEN5 software.
A user-script is then created to import the contents of the data file and store the results into a matrix of variables representing each well. A Boolean variable, which is initially set to false, is then set to true when a result is found in a well with the appropriate level of activity; the PlateCrane moves the plate to the deck of the SOLO pipettor when hits are identified. The SOLO adds a reagent, the mixture is shaken, and the plate is then returned to the Synergy 4 for the new test. If a hit is not found, the variable remains false and the PlateCrane is called to remove the plate.
Example two: interactive cherry-pickingThe protocol outlined in example one can lead to increased efficiency, but only if a limited percentage of hits are found. If the density of hits is high enough, it is likely that one hit will be found in every plate, and the above procedure would then require every plate to undergo the second-generation manipulations described, which may prove cumbersome.
One way to improve upon this is to build in an automated cherry-picking step to the protocol. A simple example is given in Figure four (page 35). In this case, the user script inside the "If-Then-Else-" statement simply keeps track of the exact location, and number of each hit. A second process is added which is designed to start when the hit count reaches the number needed to fill a new plate. The process then searches the stacks for the plates containing the hits, and systematically builds a new plate, along with the appropriate amount of serial dilution for computing dose-response curves.
Example three: Assay designThis last example demonstrates how a researcher might combine the type of "If-Then-Else-" logic described in example two, with a "While" statement to systematically modify the conditions of an assay step until the best results have been achieved.
In this example, the process begins with a Boolean variable set to false. We then hit the beginning of a "While…While-End" loop set to run until the Boolean is true. A user script is then employed (Figure 5) to extract and analyze the results from the fluorescence reader. The conditional statement that follows checks the fluorescence intensity, but this time, an unsatisfactory result leads to the addition of additional reagent and the sample is re-submitted to the multimode reader. This process continues until a satisfactory result is obtained and the Boolean is changed to true and the While-End step can be passed. One can optimize a variety of assay conditions by introducing a similar loop for each condition of interest.
ConclusionThere are many additional ways in which software can be used to automate complex multi-step laboratory protocols. Similar approaches can be applied to turn the molecular biology workcell shown in previous examples into a series of virtual workcells capable of preparing and producing proteins and nucleic acids. The two instrument groups can even be combined to form a complete assay development system that can support all stages of preclinical research. Hopefully this discussion will spark ideas into ways in which lab automation software can help the reader get the most out of their particular group of instrumentation.n
Alan H. Katz is the Chief Scientific Officer at Hudson Control Group in Springfield, New Jersey. He was previously a Principal Research Scientist in Medicinal and Computational Chemistry at Wyeth Research. Alan graduated from the State University of New York at Buffalo with a BA in chemistry and Princeton University with a PhD in synthetic organic chemistry.