Why Automation Will be the Cornerstone of the Post-COVID Biology Lab
Why Automation Will be the Cornerstone of the Post-COVID Biology Lab
Complete the form below and we will email you a PDF version of "Why Automation Will be the Cornerstone of the Post-COVID Biology Lab"
The 21st century has seen a revolution in how we approach research. With the advent of new technologies like automation and machine/deep learning, biology is becoming a less inferential and more data-driven discipline. With scientific research delayed worldwide due to the COVID-19 pandemic, is it time to re-examine the future role of automation in biological research?
The COVID-19 challenge
Many labs not working directly on COVID-19 were forced to shut down during the lockdown phases of the pandemic response, frustrating many researchers’ work. Synthace, a scientific software company based in London, were also forced to close their own lab but continued to aid research labs still operating via their cloud-based Antha platform, which helps automate biological experiments.
“We have developed new ways of deploying our software to scientists remotely. It wasn’t seamless at first but there is an incredible sense of unity at the moment. With their help and patience, we have a great new system for getting new projects kicked off halfway across the globe,” explains Tim Fell, CEO of Synthace.
Labs carrying out essential research and diagnostics for COVID-19 quickly leveraged the power of automation to increase capacity. Tecan is a global provider of laboratory instruments specializing in the development, production, and distribution of automation solutions for laboratories in the life sciences sector. They have been involved from the start, supplying local in vitro diagnostics companies in China, to providing instruments and consumables to diagnostics companies in the West. In the UK, Tecan supported the COVID-19 mega lab in Milton Keynes, providing new instruments and repurposing equipment borrowed from universities across the country.
“On the clinical lab side, we saw a tremendous need to scale up rapidly with automation solutions. Some labs who didn't yet have automation wanted to immediately scale up to several liquid handling systems to process thousands of samples per day where before they only processed a hundred samples per day,” says Luca Valeggia, who leads the Laboratory Automation business within Tecan’s Life Sciences Division.
Evolving the laboratory
While diagnostics and drug development seem best primed to benefit from automation and digitization of manual tasks, the impact of these technologies goes much further. “Basic research and discovery are expansive, explorative activities relying on a multitude of methods and a plethora of equipment. Development on the other hand is more of an optimization activity. The scientists and engineers there are generally already familiar with statistical methods of optimization and have made investments into automation but are often frustrated by the difficulty in deploying it,” says Fell.
“Even if the automation is already in place, the question is how can we better enable the remote design of experiments, remote execution, and remote monitoring of what's happening in the lab?” adds Valeggia. “Organizations need to adapt to become more and more flexible. We need to collaborate in a new way using digital tools, in and out of the lab.”
Tecan have already developed software monitoring applications to augment their devices. Their Introspect™ app is a cloud-based system which lets the user track instrument and consumable usage, while the Connect™ app allows remote interfacing via a mobile device. Modern interfaces like this could be the gateway to getting more researchers involved in using automation and cloud-based tools.
“It's all about those seamless interfaces between the equipment, software and reagents – that's the ultimate goal. We're all so familiar with our phones, and there's so much technology in there working together seamlessly. There's an expectation now, certainly from the younger generation, for software to be able to smooth over any gaps,” explains Fell. “They’re asking: why doesn't it simply all work together?”
To this end, Synthace and Tecan have partnered to use Antha to translate experimental designs in everyday language into machine-readable instructions for Tecan’s liquid handling and analytical equipment. This is followed by data handling operations, which are performed by Antha. The first application that will be jointly launched is for high-throughput, automated protein purification.
Scientists are aware that automation can achieve higher throughput, minimize human error, and ultimately increase productivity by freeing up the researcher’s time. The main roadblocks to adopting automation, in addition to considerable cost factors, are often mental barriers.
“There can be a fear that you need to be an automation expert in order to deploy and use it, and that even when you have those skills, it still takes significant time to re-program it for each new experiment. That leaves people reaching for their hand-held pipette, which in turn anchors their thinking to small-scale, one-factor-at-a-time experimentation. Breaking that ideology, which bioscience R&D has become locked into, is the most important barrier to overcome in my opinion,” explains Fell.
From Synthace’s perspective, R&D in biology suffers from tunnel vision: a single-factorial approach to experimentation. “A wider understanding of multivariate methods of experimentation would help a lot in breaking away from the current mindset, which frankly is holding biology back,” says Fell.
Valeggia notes that lab-wide links to the cloud are another bone of contention in the research community around software adoption and digitalization.
“Although many of us have transitioned to cloud-based software across our personal and professional lives, the same cannot be said of research teams. There’s still often a fear to connect the laboratory to the cloud,” says Valeggia. “Ultimately, scientists need to see clear benefits over the risks. We believe there’s a clear benefit in being able to access the lab from anywhere to see what's happening and capture insights. The resistance is decreasing as more and more mature applications become available on the market, and users experience the benefits of having new software versions released seamlessly to their devices.” he adds.
In the past decade, life science researchers have faced a reckoning of the true scale of their reproducibility problem, with significant negative impacts on healthcare, as well as wasted resources and public funding. An infamous Nature survey in 2016 found that 70% of researchers could not replicate the findings of others, and 60% could not reproduce their own findings. At the same time, automation design has become more and more flexible and user-friendly. So how should organizations now go about increasing their use of automation?
Valeggia highlights the importance of really understanding your own needs by breaking down your thinking into the following:
- Goals: What do you want to achieve? Do you have a very focused use case, or are you aiming for flexibility? Automated systems can always be readjusted or repurposed.
- Redundancy: If a robot needs maintenance, would you be able to continue your work manually? Or would you need a second robot?
- Data: What systems do you currently have in place? Do you need to plug into or update your electronic lab notebook (ELN) or laboratory information management system (LIMS)? Most importantly, what insights are you looking to get out of your data?
- Integration: What is upstream and downstream of your instrumentation? How do you want to work in the future? This is a good opportunity to consider digital connectivity for instrument monitoring in addition to data integrations.
Fell adds one more key angle that must be looked at early on:
- Optimization: How will you optimize new and existing protocols on your automated equipment? Have you considered using design of experiments (DOE) to accelerate your method development?
“If you’re starting out on your automation journey, and you're unfamiliar with DOE, I’d really recommend looking up the Wikipedia page for a quick primer. On multiple occasions, I've seen skeptics turn into evangelists in a single experiment. Once someone has gotten started, they’ll be clamoring for more flexible automation and structured data at scale,” stresses Fell.
Biological research is already seeing an abundance of new technologies leveraged in the fight against COVID-19. Automated liquid handlers are performing precision experiments to inform and characterize new therapeutic candidates as well as millions of diagnostic tests on a daily basis. Machine learning platforms are accelerating data analysis and predicting treatments for investigation.
“But I have a strong caveat to that,” says Fell. “These machine learning tools are only as powerful as the data they are given to work with. Raw data alone is meaningless. A string of numbers like 3, 4, 7, 12, 23 could be anything – a radio signal from space, cyclists on a bike path or a measure for fluorescence from a plate reader. It is the metadata that gives it context and meaning, and much too often we collect far too little metadata in our lab notebooks – electronic or paper – for these learning tools to work their magic.”
“This places a huge responsibility on the data generator, that researcher in the lab, as if they fail to collect and associate sufficient metadata with the raw data at the time of the experiment that opportunity is lost,” Fell continues. “Hence, the most important thing that has to happen in biological R&D is an across the board move from manual to automated, in all its forms – physical execution and digital data collection and structuring.”
There are now a number of opportunities to accelerate digitization to enable researchers to work in a totally different way.
“Thinking about augmented and virtual reality, we can start helping scientists to visualize what they will purchase and how it will work when it’s installed in their lab – that’s become more important now that we have less access to labs,” says Valeggia.
“We can also make our equipment even more intelligent,” he continues. “You can now have more than 100 sensors on a robot collecting metadata. Even with older instruments, some monitoring of environment changes can add valuable information. The next level is coupling all that with AI to quickly create insights for scientists into what happened on their runs.”
“The third key area is service digitization. This is a big topic for equipment manufacturers because we still need to service instruments physically. We’re actively seeking new ways to deliver more uptime to customers through preventative and remote servicing,” Valeggia adds.
Research labs, both in industry and academia, need a 21st century upgrade. The COVID-19 crisis has shown how this is within reach, as labs pivoted swiftly into diagnostic powerhouses with many integrating existing equipment into their new workflows. The robots themselves are inherently flexible: today a diagnostic tool, tomorrow a research tool and vice versa. With integrative software making such instruments simpler to operate and reprogram, researchers can now take the matter of making flexible use of automation into their own hands. It is not a question of whether digitization will be integrated into the lab – the real question is how soon?
About the authors
Tim Fell is a former Chairman of the UK BioIndustry Association’s Engineering Biology Advisory Committee and CEO of Synthace, a software company accelerating biological discovery and optimization through its cloud software platform, Antha [Updated, August 27, 2020] .
Luca Valeggia leads the Laboratory Automation business within the Life Sciences Division of Tecan, a global provider of laboratory instruments and solutions in biopharma, forensics, and clinical diagnostics.