In an effort to increase the return on investment of research and development, pharmaceutical companies are constantly looking for ways to reduce costs. One way to do this is to incorporate technology and minimize human interaction. Automation of routine experiments has increased throughput and reduced the manual steps involved. This trend toward automation is fully expected to grow over the coming years, as more lab work is conducted using advanced technologies. As a direct result, scientists will reduce the time spent in the actual lab, allowing them to focus on the design and analysis of experiments. The future laboratory will need to reflect this change, with a strong focus on automation.

The four major areas of automation as it pertains to drug discovery are: target identification and validation, assay development and screening, lead compound identification and optimization, and preclinical studies (in vitro/in vivo). Although not comprehensive, these four areas represent a significant portion of the overall drug discovery process. The demands of increased throughput, shortened time lines, and reduced resources have already led to some automation in these key areas. The full extent of automation in R&D is yet to be seen. But several areas, particularly screening, are already significantly automated.


When it comes to the Lab of the Future, there are four main areas of automation: microfluidics/lab on a chip, artificial intelligence and machine learning (AI/ML), liquid handling robotics, and data workflows.

The first category, microfluidics/lab on a chip is about miniaturizing and integrating multiple steps or functions of an experiment. Typically, these chips can handle extremely small fluid volumes, down to less than picoliters. There are several examples of these, many of which use microfluidics to direct, mix, separate, or manipulate the fluids.

The second area is the use of AI/ML to accelerate various aspects of experimentation. One such example is active machine learning in drug discovery, which can not only predict screening results, but also develop models to efficiently select which experiments to perform.

The third area is liquid handling robotics, which essentially dispense a selected quantity of reagent, sample, or other liquids to a designated container. As automation continues to get easier to use, many platforms (providers) make it possible for scientists to create and share customized workflows, which also increases collaboration.

The fourth and final area of automation is data workflows. There are many examples of Internet of Things–connected instruments, which send data to the cloud automatically. Often, these cloud solutions have hosted pipelines for data analysis, which is also performed in the cloud. These analyses can range from basic yes/no types of answers from a PCR experiment to sophisticated secondary and tertiary analyses of next-generation sequencing (NGS) data. As described in the active machine learning article referenced above, the output of one experiment can be the input of the next experiment. When fully integrated, this can result in an endless series of experiments, continuously optimizing the model.


There are, of course, many challenges and difficulties, particularly in implementation. One such challenge is interdisciplinary coordination. With an increased use of robots and automation, how do you address the need for greater teamwork across disciplines? If researchers are no longer interacting in the lab itself, they need to find other ways to communicate and collaborate. Also, integrating legacy instruments into an automated lab workflow presents several obstacles. Often older instruments do not have modern methods of connectivity, which can make integrations difficult, if not impossible. Finally, standardizing formats will be important, particularly for aspects of the Lab of the Future that support automation, such as incorporation of data gathered from Internet of Things devices and sensors, which will come in a variety of different formats and structures. Addressing these and other challenges is difficult and will require input from multiple stakeholders.

Looking Forward

The Lab of the Future is just that, a forward-looking vision of research labs. Many labs, both commercial (biopharma, contract research organizations (CRO), etc.) and academic, have already been centralized using a “core” model. Over time, this trend of highly specialized core facilities that focus on a limited number of areas will increase. Additionally, the manual steps will decrease, and the physical act of carrying out experiments will lean more heavily on various forms of robotics. This will not only decrease costs, but also increase reproducibility and throughput. Finally, there will certainly be a key role for AI in augmenting lab automation, not only in the design of experiment phase of selecting the highest value experiments, but also in the analysis and interpretation of data. It is reasonable to expect automation to play a huge role in the LoTF, but there are limits. The role of scientists in the future will most certainly change and evolve over time. However, there are certain aspects that cannot be automated, particularly around the scientific method: being curious asking questions, examining information and evidence, and figuring out conclusions.

Listen to the podcast.

Listen on Apple Podcasts

Previous post

Active Machine Learning in Drug Discovery

Next post

Improving CRISPR-Cas9 Gene Editing:
Tackling the HDR Inefficiency Problem