Month: June 2021

Frequently, there are very repetitive and laborious tasks in a laboratory that could be automated. Whether it’s compiling data on a spreadsheet or preparing a gel for electrophoresis, the ratio Value generated / Time used is not often high. Derek Lowe, a chemist and author of the Into The Pipeline blog (1), humorously recalls a time when a simple chromatography took an enormous amount of time to perform (a time now almost over), and correctly notes that the goal of automation is not to push the researcher out of the lab, but to reduce all those laborious tasks and support the scientist’s intellectual input to the maximum.

In Chemistry or Biology, many groups are trying to imagine the laboratory of the future, one that could carry out end-to-end the synthesis and the testing of a molecule. However, from a technical point of view, the range of actions required to reproduce the work of a researcher by a robot is far too wide to be effective today, but the various projects presented below are promising for the future.

***

In the field of diagnostics, it is becoming increasingly vital to automate tests, in order to meet the demands of patients and clinicians. To give you an idea, in the UK, almost one million PCR tests are performed every day for covid-19 alone (2).

In clinical microbiology the use of automated systems is particularly interesting, where protocols require a lot of time and attention from microbiologists. The WASP robot, designed by Copan (3), combines robotics and software and is capable of performing culture operations, bacterial isolation, and monitoring whether the growth is done correctly thanks to a small camera installed in the robot. There is also Roche’s Cobas (4), which is capable of performing various molecular biology tests such as qPCR. The versatility of these robots allow them to be easily adapted for other diagnostic purposes.

***

In a more chemical context, a Liverpudlian group led by Prof. Andrew Cooper (5) has designed a robot capable of physically moving around the laboratory in order to optimise hydrogen production by photocatalysis. The advantage of this robot is that it is human-sized and can operate and move freely in any room. Although it took some time to set up such a system, it is estimated that the robot was 1000 times faster than if the work was done manually. The video of the robot is available below:

The artificial intelligence implemented in most robots works by iterations: the results of each experiment are evaluated by the algorithm and allow it to design the next experiment.

Figure 1: The combination of Artificial Intelligence and robotics allows the creation of an iterative circuit, where each cycle analyses the results of the previous one, and adapts the parameters to optimise the process defined by the researcher

In the field of Materials Chemistry, Alán Aspuru-Guzik et al. (6) have developed an automated and autonomous platform capable of working with a large number of parameters in order to discover new materials, useful for solar panels or electronic consumables. In Organic Chemistry, Coley et al (7) have used AI and robotics to synthesize small molecules by Flow Chemistry. All the chemist has to do is to indicate the molecule one want to obtain, and the AI will carry out its own retro synthesis pathway and try to synthesize the compound. The automaton was able to synthesise 15 small therapeutic molecules, ranging from aspirin to warfarin.

Other initiatives can be noted, particularly from Big Pharmas, such as AstraZeneca and its iLab (8), which aims to automate the discovery of therapeutic molecules via an iterative circuit of Design, Make, Test, Analyse. In Medicinal Chemistry, combinatorial chemistry methods enable the chemical space of a target to be explored very rapidly, thanks to controlled and optimised reactions. These projects witness the progress towards totally autonomous synthesis systems.

***

It is probably fair to note that some researchers are wary of using robots, and sometimes feel threatened by being replaced by a machine. Myself, as an apprentice chemist and exploring the subject, have said to myself on several occasions “Hey, but this robot could work for me! I remember the weeks I would spend trying to optimise a reaction, chain testing different catalysts, a job that an automaton (or a monkey!) could have done for me much more quickly and certainly more efficiently. Robotics has this enormous potential to improve the productivity of researchers, and to reduce tedious tasks that ultimately require little or even no intellectual thought at all.

There are also tools that help the researcher to develop research designs to optimise an X or Y process in the most efficient way. For example, the EDA tool developed by NC3Rs (9) is useful for in vivo research projects, where one tries to obtain statistically powerful data while reducing the number of animals used. Other tools have also been developed using Bayensian or Montecarlo Research Tree models (10), and lead to optimal experimental designs. In the same vein, Aldeghi et al. have developed Golem (11), an open-source tool available on GitHub (12).

Cloud technologies (i.e. access to a service via the internet) also hold great promise for the laboratory of the future. They will allow researchers to carry out their research entirely from home, thanks to “a few” lines of code. Projects such as Strateos have initiated this practice and already allow researchers to program Chemistry, Biochemistry and Biology experiments from home. Once the protocol is defined, the researcher simply launches the experiment from one’s computer and the robot located thousands of kilometres away carries out the operation for one’s. In a few years’ time, if the service becomes more widely adopted in the scientific community, everyone will have easy access to it for a reasonable price.

Figure 2: Cloub Lab principle. 1) The researcher sends his or her research protocol to the automaton, located on the other side of the world. 2) The automaton carries out the experiment designed by the researcher and 3) returns the results to the researcher as soon as the experiment is completed.

***

Between progress and doubts, it is probably only a matter of time before the scientific community adopts a different mentality. There was a time when telephone switchboards were run entirely by people, until the day everything was replaced by automatons. This short documentary by David Hoffman crystallised this transition and the reaction of users when they hear a robotic voice for the very first time. Although some of them were reluctant at first, the implementation of voice recognition has made the service much more efficient and less expensive for consumers. Won’t tomorrow’s researchers all be roboticists to some extent?

Bibliography

1. Lab! Of! The! Future! | In the Pipeline [Internet]. 2021 [cited 2021 Jun 9]. Available from: //blogs-sciencemag-org.ressources-electroniques.univ-lille.fr/pipeline/archives/2021/03/31/lab-of-the-future
2. Testing in the UK | Coronavirus in the UK [Internet]. [cited 2021 Jun 13]. Available from: https://coronavirus.data.gov.uk/details/testing
3. Copan WASP DT: Walk-Away Specimen Processor [Internet]. [cited 2021 Jun 9]. Available from: https://www.beckmancoulter.com/products/microbiology/copan-wasp-dt-walk-away-specimen-processor
4. Automation in Molecular Diagnostic Testing [Internet]. Diagnostics. [cited 2021 Jun 13]. Available from: https://diagnostics.roche.com/global/en/article-listing/automation-in-molecular-diagnostic-testing.html
5. Burger B, Maffettone PM, Gusev VV, Aitchison CM, Bai Y, Wang X, et al. A mobile robotic chemist. Nature. 2020 Jul;583(7815):237–41.
6. MacLeod BP, Parlane FGL, Morrissey TD, Häse F, Roch LM, Dettelbach KE, et al. Self-driving laboratory for accelerated discovery of thin-film materials. Sci Adv. 2020 May 1;6(20):eaaz8867.
7. Coley CW, Thomas DA, Lummiss JAM, Jaworski JN, Breen CP, Schultz V, et al. A robotic platform for flow synthesis of organic compounds informed by AI planning. Science [Internet]. 2019 Aug 9 [cited 2021 Jun 3];365(6453). Available from: http://science.sciencemag.org/content/365/6453/eaax1566
8. The AstraZeneca iLab [Internet]. [cited 2021 Jun 27]. Available from: https://www.astrazeneca.com/r-d/our-technologies/ilab.html
9. du Sert NP, Bamsey I, Bate ST, Berdoy M, Clark RA, Cuthill IC, et al. The Experimental Design Assistant. Nat Methods. 2017 Nov;14(11):1024–5.
10. Dieb TM, Tsuda K. Machine Learning-Based Experimental Design in Materials Science. In: Tanaka I, editor. Nanoinformatics [Internet]. Singapore: Springer; 2018 [cited 2021 Jun 6]. p. 65–74. Available from: https://doi.org/10.1007/978-981-10-7617-6_4
11. Aldeghi M, Häse F, Hickman RJ, Tamblyn I, Aspuru-Guzik A. Golem: An algorithm for robust experiment and process optimization. ArXiv210303716 Phys [Internet]. 2021 Mar 5 [cited 2021 Jun 9]; Available from: http://arxiv.org/abs/2103.03716
12. aspuru-guzik-group/golem [Internet]. Aspuru-Guzik group repo; 2021 [cited 2021 Jun 9]. Available from: https://github.com/aspuru-guzik-group/golem