Robotics

IPT talks to Roya Amini-Naieni at Trilobio about robotic automation in pharma research, the different applications of soft-and hardware, and how the field is likely to develop over the next five years

IPT: What was the landscape of biologics research like before robotic automation?

Roya Amini-Naieni (RAN): The inconvenient truth is that the landscape of biologics research (as distinct from high-throughput manufacturing and production) is much the same today as it was before the rise of robotic automation. Today, large, well-capitalised companies can afford robust automation for their high-throughput production level protocols because those protocols are usually profitable, highly optimised and unlikely to change on a regular basis. This automation is very expensive because of the scale and sophistication of the production process. of a single procedure in a lab, but the massive tasks in R&D labs are mostly not automated. The experience of these labs is similar to what it was like before automation; they fall back on manual processes without having their painpoints addressed. Without widespread, walk-away automation, many scientists are working hours each day on repetitive tasks that do not leverage the unique capabilities of a human (like creativity, analysis or problem-solving). These tasks include colony picking, in which a human looks at a petri dish, counts and circles colony dots from hundreds of dots on a petri dish, and then picks some of them with a pipette tip. PhD students, professors and CEOs of start-ups are doing this work, which should be done by automated technologies.

In contrast, R&D labs mostly have no significant automation and conduct much of their exploration manually. The field is still very much the same – automation is expensive and so it can only be fully justified in high-throughput production at large organisations that can afford it, effectively skipping over the R&D use case. The problematic result of this is a reproducibility crisis; 77% of biologists cannot reproduce their own or other’s research.1

Some may think of a polymerase chain reaction (PCR) machine as automation because scientists used to have to manually transfer samples from different water baths to cycle temperatures. In a sense, this is a kind of automation

IPT: How are automated robotic lab systems improving research inefficiencies?

RAN: There are three types of inefficiencies to discuss:

Functional
Biologists are measured against how consistently they can perform their research. This can lead staff to feel as if they are trying to be a robot instead of a person, and that the only sign of a good lab associate is one who can do the same thing over and over without mistake. However, these are not the tasks where humans excel, and so there are various inefficiencies in throughput, accuracy, precision and man-hour limitations. As we know, lab automation addresses these inefficiencies by reducing variance in the execution of the R&D protocols and increasing accuracy. Furthermore, automating these tasks and unlocking the potential for 24/7/365 walk-away automation can unlock billions of hours of research each year and major biological breakthroughs.

Opportunity cost
There are only so many hours in the day, and a biologist can only physically work on one thing at a time. Manual work takes biologists away from other, and potentially higher value, research, development and analysis activities. On the other hand, if a biologist isn’t performing manual research tasks, they can instead be designing a new experiment, analysing results in a new way or collaborating with other biologists. advanced functionality with the software. For example, adding new and more sensors on a lab device has the potential to automate data collection and lab workflows, something that is not otherwise possible. In another example, hardware and software can abstract away plasticware so biologists no longer need to manage specific wells and their contents during the R&D process, shifting the focus from tediously managing these details to the design and modification of an experiment itself.

A horizontal software layer that spans across and orchestrates as many devices as possible across the lab is incredibly important. In other words, we should not be building lab automation software in the future that is stuck within a single device and accessed through a small liquid crystal display (LCD) control panel and thumb drive.

Psychological
Common worries for staff include remembering if they moved a sample to another tube, or if some material was moved into the wrong well in a 96-well plate; among all the tasks performed that day, did they make just one mistake that messed up the research? Robotics solves this problem 100% by eliminating the potential for human error. Scientists are trained to think creatively about biology, but instead are worrying about contaminating their samples or moving things to the wrong place instead of focusing on inventing new drugs or experiments.

IPT: What devices are integral to robotic lab systems, and why is it important these devices are automated?

RAN: Though it may seem too ambitious, the dream is full automation. Each device that a biologist needs to complete their research is integral. The way that biology can achieve its true potential is by mimicking computers – biologists should be controlling their labs like a computer scientist controls their computer. Every piece of equipment should be designed for automation and that includes everything in the lab. This means we have to start thinking differently about how these devices all work together better. The next step in expanding lab automation is to connect devices instantly and without complex code or calibration. This allows the field to begin to approach whole lab automation.

IPT: What is the division of hardware versus software in lab automation, and how does one support the other?

RAN: There should be no division between the hardware and software in lab automation. Ideally, the power of the software should be enabled by the innovations of the hardware. At the most basic level, high-performance, reliable hardware is a rate limiting factor for the software, but it is easy to get really excited about the game-changing software functionality enabled by new hardware paradigms. Lab automation can solve unique problems with innovative hardware, and then expose and operationalise that

IPT: How will the integration of automated robotic lab systems develop over the next five years?

RAN: The industry will see a shift towards designing automation specifically designed for the R&D use case, which has significantly different requirements than high-throughput manufacturing and production. This will make R&D lab automation more accessible and scalable, unlocking billions of hours of research time and the immense potential of these biologists to solve a larger variety of problems for the world. The hope is that scientists and biologists will look back in ten years and think it was crazy they had to manually pipette at all, or most manual research tasks for that matter. The lab will become a cube, or collection of cubes, that a biologist can control from their laptop. Automation will enable a single biologist to be able to do the work of hundreds of biologists, unlocking a huge amount of research and data. It will free up more time for creative endeavours and empower biologists, rather than the current state of the lab, which is quite burdensome.

Reference:
1. Visit: https://www.nature.com/articles/533452a

Roya Amini-Naieni is the CEO and co-founder of Trilobio, a robotics and biology research platform with the mission to fully automate biology for everyone. Trilobio was founded in 2021 and has raised over $3m to date. Roya studied Mathematics and Computational Biology at Harvey Mudd College (HMC), US, where she founded a SynBio lab with grant funding from HMC’s president. Previously, she conducted SynBio research at the University of Washington, US, HMC and at Asimov. Roya also founded the first non-collegiate iGEM team in Washington state. She is a 776 Fellow and Forbes 30 Under 30 recipient.