Listen to this article on the DDW Podcast:
Joshua Kangas, an educator in the Computational Biology Department of Carnegie Mellon University, is helping to shape the minds of future lab specialists. He tells Megan Thomas what he believes the lab of the future will look like in this exclusive DDW & SLAS2022 supplement ahead of SLAS2022 in Boston, US on 5-9 February.
Dr Kangas teaches laboratory courses, where students learn about experimental design and execution as well as the computational skills needed to analyse the resulting data.
“A significant amount of my effort goes into the MS Automated Science programme. In this programme, students are trained to use artificial intelligence methods to direct lab automation robots to run experiments. I also manage the Automation Lab and maximise availability of the system for research efforts when it is not being used for coursework,” he says.
Perks of the job
Kangas adds: “Like many in an academic setting, I get great satisfaction out of the moments working with students when the ‘light bulb turns on’. In my classes, I most frequently see these moments when the students are working on a challenging robotic technique and they finally solve a problem they’ve been working on for hours. Or, sometimes, they’ll be working on an algorithm to analyse data in a certain way and they’ll finish a successful implementation of it and really clearly understand how it works.
“Perhaps my favourite examples of these moments are when students are able to draw a clear connection between the way they set up and ran their experiments and the way they performed the computational analysis. As a bit of a hardware geek, I also really enjoy working with robotics to accomplish new and interesting tasks.”
“One of the major challenges facing educators working in science labs during the pandemic was the task of safely offering laboratory classes which require students to be physically present in the lab,” Kangas says.
“Our Automation Lab was designed to be run with minimal human intervention and as soon as the pandemic started, we immediately installed cameras inside the robot enclosure and set up our lab computers for remote access. These two steps allowed students to remotely log in to the systems controlling the robot to design and execute experimental protocols.”
Then, they could log into the system controlling the cameras and watch the robot run their experiments from wherever they were in the world. “Although there were some challenges, these lab automation courses were able to continue with minimal disruption through the pandemic. Now that courses are back in person, I have used some of the same techniques in person,” Kangas says.
Pandemic and lab automation
“It should be clear from the efforts for Covid testing that lab automation can be quickly leveraged to address enormous problems rapidly,” Kangas comments. “Within research labs, I think many more people are starting to think about the potential value in automation to allow people to do their work remotely. Furthermore, the pandemic has brought to light efforts by some companies to offer remote access to lab tools.”
AI: Efficiency, speed and cost
Kangas believes that lab automation has often been viewed as allowing experiments to be run faster, smaller, cheaper, and more reproducibly. “Depending on the application, these are more or less true in drug discovery, however lab automation hasn’t resulted in a boom in new drug discoveries,” he says.
“My view is that the next major advancement will be to use AI to control lab automation in closed-loop experimental processes to answer complex questions efficiently with a minimal number of experiments executed. Modern AI methods allow for the selection of the most informative experiments and by focusing lab automation on running the informative experiments, overall substantially less experimentation is required to build accurate predictive models. This means that researchers can tackle far more complex problems with a reasonable amount of experimentation when driven by AI methods.”
Thinking about the challenges and opportunities ahead, Kangas says: “From my perspective, one of the major challenges is identifying and hiring people with the skillset necessary to properly leverage automation and AI together. I am increasingly hearing about companies who are interested in these skillsets facing struggles filling these roles. I hope my teaching efforts can be a part of the solution to this challenge. One of the opportunities I see is in the use of machine learning methods to pull data from disparate sources together into accurate models of biological phenomena. Some efforts have been made in this area of sharing data across siloes through various consortia, but the careful application of machine learning/AI to this problem may yield great benefit for involved organisations.”
The lab of the future
“I envision that a substantial amount of work will be done by robots in remote facilities in the future. Lab automation and AI together will change the way many researchers operate. They will need to think less about how to design and execute experiments. They will spend more time thinking about the larger questions to answer and how to measure biological phenomena related to those questions. Then AI can be used to drive experimentation to rapidly and efficiently provide answers.”
Volume 23, Issue 1 – Winter 2021/22 | SLAS2022 supplement
Joshua Kangas earned his Ph.D. in Computational Biology from Carnegie Mellon University. The laboratory courses he teaches are focused on the interface of computation (including machine learning and modeling), biological data generation (sequencing, microscopy, cytometry, etc), and artificial intelligence-directed laboratory automation.