Why integrated data handling is critical for successful automated cell based assays
The wealth of advantages afforded by automating cell screening assays is clear. Generating high quality, more reproducible, data yields robust experimental results, while laboratory staff are freed from the burden of manual repetitive steps to focus on more dynamic tasks. Enhancing laboratory quality and efficiency, automation is a driving force behind scientific discoveries - and is a popular solution for cell screening applications throughout pharma and academia.
Sparked by the evolution of reliable and affordable automated microscopes, the past decade
has witnessed a surge in the popularity of automated cell screening assays. A range of
biological assays can be utilised to investigate various properties, such as cell viability assays for investigating cytotoxicity. Meanwhile, non-cellular compound screening assays have become largely exhausted, with efficacy often failing when hits are tested on cells. The complexity of living cells means that the effects of a compound can be highly unpredictable and testing compounds in biologically-relevant models from the beginning saves time and resources.
Although automated cell screening platforms are mainly associated with liquid handling, there are various other factors that should be considered. For example, maintaining cell culturesin a stable and sterile environment is of paramount importance in order to protect operators and samples, and in turn increase the overall success of the assay. In addition, many scientists underestimate the significant data load generated by automated systems, which can quickly become unmanageable.
Often, due to regulatory requirements (for example, FDA approval), stored data needs to be traced back to a corresponding sample, perhaps even a number of years after the initial experiment. Therefore, it is vital that effective data handling systems are in place. Without such automated data handling, resource-intensive data analysis creates a bottleneck that undermines the efficiency benefits afforded by the automated cell screening setup.
Considering data handling as an integral part of the automated system build is therefore
key to maintaining efficiency and reaching the system’s full capacity from the outset, yet the
fundamental importance of this is often overlooked. While a build-only solution may initially
seem a cost-effective means to automation, hidden costs are later revealed when it comes
to software and data handling. This white paper explores how integrating data handling into
automated cell screening platforms delivers significant advantages – which are crucial to
consider, and detrimental when overlooked.
Why integrate data handling into automated systems?
Automated cell screening platforms generate a high data load, and while the cost of a system build including full data handling integration may initially appear higher, in the long term this saves both cost and time:
• Eases the bottleneck of data analysis, enabling full capacity of automated systems to be reached
• Standardised data analysis yields reliable results
• Full tracking makes it easy to identify each sample parameter
• Direct collaboration between supplier and laboratory IT experts achieves seamless informatics
• Ongoing upgrade support (e.g. Windows 10), future-proofing the system
• Regulatory requirements.
What to look for when automating cell screening assays
The move to automate cell assays is a logical progression from manual
operation, and efficiency improvements are significant. In the case of
antibody screening, for example, each automated system requires one
single full-time equivalent (FTE) for every four that a manual screening
campaign would typically involve.
Operators are therefore no longer required to undertake repetitive tasks, which reduces the risk of repetitive strain injury, and allows them to dedicate more time to responsibilities such as results analysis, reporting and preparing publications. In addition, the risk of human error is bypassed, which includes pipetting errors, contamination and management of cell cultures. As cells are highly sensitive to variations relating to passage timing and media changes, fluctuations caused by manual errors can subtly and unintentionally alter cellular phenotypes, which will influence downstream results. In contrast, these tasks are highly standardised with automation, which leads to accurate and precise cell screening data being acquired in a shorter time, with ease and safety.
The automated system itself is defined by the arrangement of hardware and software features, often formed from a combination of off-the-shelf modules such as robots and automated microscopes, chosen to match specific requirements. Developing the system from proven instruments ensures reliability while minimising build times, and these can then be built around a customised workcell enclosure. When handling cell cultures, the workcell must provide a sterile and highly controlled environment, avoiding further variations in cell handling and contamination of valuable cell samples - and also protecting the operator. This is especially true when adhering to regulatory and quality control standards, which demand a guaranteed aseptic environment.
Figure 1: A sterile workcell is vital for automated cell screening.
PAA’s S-CELTM provides biological class 2 containment, providing both operator and sample protection.
For this reason, systems with ISO-certified laminar flow setups provide the confidence that
operations will work at the same performance levels as a conventional clean hood ( Figure 1). Across the workcell, the separate hardware instruments are linked together with scheduling software, which both releases and retrieves data.
1) Flexible data handling
Data generated by automated cell screening assays can include experimental history, numerical data and microscopy images, and this is where flexibility comes into play to provide the most comprehensive and insightful analysis. For example, numerical evidence of cell viability is supported by the associated microscopy image and signal quantification protocol applied. It is therefore valuable to store and analyse the relevant data in parallel, and adapting this process to the existing data infrastructure ensures seamless operation within the laboratory workflow. Enabling database communication and data handling in the desired format, flexible scheduling software such as PAA’s OverlordTM can be tailored to each system via purpose built plugins.
Links can be made to an internal laboratory information management system (LIMS) for sample identification, but also to third party data analysis programs such as Spotfire® (TIBCO, Palo Alto, USA) and ActivityBase® (IDBS, Guildford, UK). Widening the scope of assays that can be fully managed, data is fed into these programs and results automatically returned based on user-specified parameters. Such versatility allows the system to be utilised for highly specific requirements, from measuring cytotoxicity via lactate dehydrogenase (LDH) release, through to quantifying each signal in multiplex luminescent and fluorescent assays. Moreover, this is achieved with minimal operator input between entering the samples and retrieving insightful
2) Easy data retrieval with sample tracking
In addition to the communication aspects of data handling, as cell samples are processed through the system, database tracking functionality ensures that a full history of each sample is readily available to the operator. This can serve many purposes, such as tracing the origin of outliers when the results are reviewed. This capability also facilitates assays requiring the association of multiple plates, for example when cell culture supernatant is assayed on one plate, while the cells remain in culture on the second plate. The results from the first plate containing the supernatant must then be tracked to the corresponding cells, providing the position and barcode for easy location of the relevant sample.
It is equally important to be aware of the challenges involved in retrospectively organising such complex data acquired over months of screening, which may be for the purpose of preparing research publications or patent applications. To place this into context, consider publishing a research paper describing past screening campaigns covering nine years of work, where five different technicians were involved in data handling. It can easily require a month to organise the data, retrieving it from multiple network locations and unformatted spreadsheets. Some experiments might even need repeating where results become lost over this length of time, and each file must be re-formatted to provide directly comparable data across the entire screening campaign. With data management systems in place, data is easily accessible and consistently formatted to streamline the reporting and publication process.
3) Operational simplicity
In essence, fully integrated systems with data handling provide a step up from physical
automation alone, efficiently managing the complete assay from start to finish. When managing assays scheduled across multiple days, the results generated at one step can dictate how the programme progresses. For example, to confirm that results are as expected and assay conditions such as reagent stocks are sufficient for the following procedure. It is essential that the operator is provided with the opportunity to check the system each morning, which notifies and guides them through each new step to be launched. Once the assay is completed and data automatically analysed, results are generated and the system provides the location of the hits, moving them directly into storage as required.
Despite the underlying complexity of such data handling and sample tracking functions, an
automated system with the correct interface is a highly accessible platform for every operator. A graphic and intuitive interface requires little time to train a new user for standard day-to-day running of every cell process on the system (Figure 2), while expert operators remain on-hand for advanced tasks.
Figure 2: An intuitive interface simplifies complex data handling.
Despite automated data handling involving a multitude of highly complex processes, PAA’s OverlordTM and
HarmonyTM provide an easy-to-use interface, streamlining daily operation even for those with minimal training.
4) Adapting to new assays with ongoing support
By choosing a fully integrated system, customer service also includes ongoing software support for assays, with remote systems providing a fast response time. PAA’s platforms run on the latest software, and the company ensures that continued support is provided, including upgrades to Windows 10. The complete system can also be upgraded to meet changing requirements, for example, when moving from adherent to suspension cell cultures. Addressing both the associated hardware and software upgrades in parallel, provides both a cost effective and efficient alternative to separating the automated build and data handling systems.
Case Studies
Cell culture assays often run over an extended time period of eight days or more, and automated systems tracking each and every sample generate a significant data load. Providing the extra layers of automated data handling necessary to manage such operations is facilitated by suppliers also offering software expertise. The following case studies highlight how the benefits of automation stem not just from the build itself, but also from
the usability and data automation behind it.
1) Advancing antibody discovery at UCB
As an alternative to the traditional low-throughput hybridoma approach, in 2013 PAA worked
with UCB to develop a novel automated platform to enhance the discovery of high-quality
monoclonal antibodies for therapeutic use.
The automated system itself is built around three integrated PAA S-CELTM class 2 containment workcells, with made-to-measure dimensions. Adhering to the relevant ISO standards, the laminar flow setup maintains a sterile environment to protect against contamination risk, and the workcells manage separate processes as follows:
1. Cell culture – In the filling workcell, B-cells collected from samples immunised with the target of interest are seeded into barcoded 96-well plates, with a typical experiment involving more than 500 cell plates. Over seven days of incubation, B-cell clonal expansion and differentiation into antibody-secreting cells occurs, and antibody is secreted into the supernatant. Cell plates are then placed on a conveyor link to the Screening Workcell.
2. Antibody screening – In a second workcell dedicated to antibody screening, secreted
antibodies on the cell plates are assayed for binding and functional activity. The supernatant
is first screened for antigen-specific antibody, with 10 μl of supernatant removed from the cell
culture plate into a 384-well assay plate. A homogenous fluorescence binding assay measures specific binding through reading fluorescence intensity, with an intensity of 10% or less in a single well indicating that the signal derives from only one B-cell clone.
Sample identification data from the original cell plates is tracked through the system
and linked with the assay plate. Following data acquisition with a high content screening
instrument, the PAA data handling module will take the CSV data files and through
ActivityBase, analyse the data against an operator-defined threshold. From this, a plate map
of positive hits is produced for the user to review and validate, forming a pick list for the Hit
Picking Workcell.
3. Hitpicking – A third workcell is optimised for hitpicking, and supernatant from the positive hits is transferred from the cell plates to a single consolidated master plate. Further analysis with complex functional and biochemical assays enables UCB to characterise the antibody in more detail. In parallel, cell plates containing the positive hits are stored at -80°C.
Full sample traceability through the system is provided, and data linking the master plate
to the cell plates is stored within a custom company database, hosted within both Oracle
and IDBS ActivityBase software. Once hits that match a panel of requirements are identified
by subsequent screening, the B-cell clones of interest are isolated for cloning and further
development using UCB’s novel fluorescence foci method (more information is available in
Tickle et al.)
Data handling is fully integrated, with PAA’s Overlord software interfacing with a web service
for continual communication with the internal database, updating all results in real time.
This includes a link to internal LIMS, as well as Spotfire® and ActivityBase® (Figure 3), while
the Harmony touch screen interface guides the operator through a series of questions to
programme each process. In order to achieve this complex network setup, PAA’s software
engineer collaborated directly with UCB’s IT department. This specialised co-ordination
enabled the seamless integration of the automated platform in terms of data format and code.
Figure 3: Data handling with UCB’s automated antibody discovery platform.
Data handling centres on UCB’s data repository web service, with links to databases and third party data analysis programmes. This functions for each of the three workcells: Filling workcell (cell plating) > screening workcell > hit picking workcell, and is managed by PAA’s Overlord software.
Increasing capacity and quality with a fully automated system
Enabling the screening of more than one billion B-cells in a single campaign, automating UCB’s antibody discovery programme with fully integrated data handling is driving the discovery of rare antibodies across a diverse array of targets and species. UCB is now able to mine the entire immune repertoire for high-quality antibodies in a shorter time, helping the company deliver efficacious therapies faster. This is achieved through a host of advantages:
• Increased capacity to 400 plates per day, allowing UCB to run more projects in parallel
• Allows antibody selection based on highly stringent criteria, generating high-quality hits
• Exact, repeatable processing yields reproducible data
• Completely sterile operation
• Intuitive operation is ideal for users with minimal training
• Scheduling provides tight control of incubation times, standardising assay
conditions for reliable results
• System has flexibility for multiple plate batches, screening multiple cell lines
against multiple compounds
• Provides a full inventory and electronic audit trail for the entire process.
2) The pitfalls of overlooking data handling
In another case, an automated cell screening platform was developed to isolate specific cell
clones, overcoming the laborious and time-consuming nature of manual plating, isolating and
screening. Through initial budget constraints, data handling costs were omitted and the system was unable to run at its full potential.
Although the aim was to run 60 plate batches, it quickly became evident that only smaller
batches of 10 plates were able to keep up with data tracking and analysis demands. These
demands also included cell confluency calculations and matching results to stored samples.
Since each batch requires several weeks for processing, smaller batches failed to make use of the automated platform’s full capacity.
Reaching full capacity with automated data handling
After weighing up the costs, data handling was retrospectively incorporated into the system,
co-ordinated by the OverlordTM software, with the HarmonyTM interface to guide user interaction. Automatic confluency measurements improve efficiency, while a bespoke system tracks clones. Multiple batches at different stages of the screening programme can now be easily selected for further processing at the beginning of each working day, and the system provides full tracking once a process is selected by the user.
Data handling also enables hits based on clone quantification and analysis to be linked to the
corresponding replicate plate in storage (see Figure 4). Providing the location and barcode, hits can therefore be easily retrieved from storage for further analysis. Streamlining data handling in this way has allowed multiple batches to run concurrently in order to maximise throughput (project duration was reduced from 12 weeks to 8 weeks, with a 15x increase in number of clones screened). This demonstrates how integrating data handling presents a cost-effective means to manage the high burden of data volume and analysis.
Figure 4: Automation of cell screening workflow and data handling.
By automating this process, each project is reduced in total duration from 12 weeks to 8 weeks, with a 15x increase in the number of clones that can be screened.
Summary
Automation of cell screening assays is popular and extremely valuable, both within the pharmaceutical industry and academia - especially in collaborative research programmes. However, as we have discussed, it is crucial that the often overlooked, yet essential aspect of data handling is given a high priority when considering a system build.
While data handling is challenging for operations of every size, it is the increased volume
of data generated by automated systems that leads to a bottleneck if not collated in an
effective way. The term data handling encompasses many tasks, from data generation
and re-formatting to suit individual requirements, through to communication with third
party software programs and databases. Full sample tracking and ease of operation is
highly advantageous for efficient routine use, and each factor must remain flexible to
fit around the great variety of demands of individual laboratories and their existing data
infrastructure.
It is little wonder, therefore, that many suppliers struggle to deliver automated systems
with fully integrated data handling, yet the importance has been highlighted by the stark
comparison between the two case studies explored. Planning for data handling from
the outset reduces project cycle times and in the case of UCB, increases the capacity to
discover high-quality antibodies. The company’s efficiently automated antibody discovery
platform provides standardised results with an intuitive interface, saving time and money
in the long-run. On the other hand, failing to plan for this highlights the downfalls of
initial cost-cutting measures in data handling, which significantly limits the capacity
of automated systems and entails additional costs in terms of time, resources and
ultimately, money.
It is clear that when automating cell screening assays, customised and fully integrated
systems must factor in not just physical automation but comprehensive data handling –
and it is vital that suppliers of automated systems respond effectively to these needs.
Comments