The Tartan

By Jimmy Baracia, Arden Ryan and Jaden Singh

Nestled in the corner of an old warehouse in Bakery Square, the Carnegie Mellon Cloud Lab is one month away from becoming the nation’s first remote academic laboratory after six years in development. Offering broad access to a host of scientific instrumentation, with over 130 unique instrument types and approximately 200 total instruments, the Carnegie Mellon Cloud Lab will enable researchers to virtually run a multitude of testing protocols.

Average labs have only five to eight instruments, said Ally Jump, assistant director of corporate and foundation relations in the Mellon College of Science (MCS), who provided The Tartan with a tour of the Bakery Square lab. Access to more instruments will open principal investigators (PIs) to a broad range of study in fields including life sciences, chemistry, computational biology, and chemical and mechanical engineering.

“I believe that the opening of the CMU Cloud Lab this spring will bring about a transformation in how we teach and do science,” said MCS dean Curtis Meyer. “We will no longer be limited by the few instruments we have available nearby, and our ability to quickly run large scale reproducible research campaigns will completely change how we think about doing science.”

The automated research facility is modeled after a startup founded by two MCS alumni. D.J. Kleinbaum and Brian Frezza built the first remote-operated laboratory, Emerald Cloud Lab, and its pharmaceutical venture, Emerald Therapeutics, after studying computational biology at Carnegie Mellon.

A team of specialists at the facility will handle all routine tasks — such as dishwashing, ordering consumables, and calibration and maintenance of the highly sensitive instruments — in tandem with a line of robots. These time-consuming tasks typically fall to graduate students, but when automated, will vastly increase productivity.

The preparation, running, and results reporting of every test will be handled centrally and sent to the PI as a data package at the end of the experiment. These data packages will include metadata such as humidity and temperature, as well as, in many cases, a video recording of the experiment, deposited in an easy-to-aggregate data file for researchers.

A major benefit of an automated laboratory is that it solves the issue of duplication in experiments, which is a notoriously difficult facet of research, Jump said. When many of the extraneous factors, such as humidity, are accounted for, and when experiments can be repeated in exactly the same way, scientific results are much more duplicatable.

The Cloud Lab is set up so that researchers around the country, and most anyone with an internet connection, can utilize the facility on a subscription basis. They will drop off their materials and handlers will work alongside the robots to move from one station to the next until the experiment is complete. Each concurrent experiment is referred to as a “thread,” usually conducted over a one-month timeframe, enabling researchers to purchase monthly rights to access the wide array of instruments at the Cloud Lab. This will lower the cost barrier to run experiments. Scientists won’t need to purchase every expensive device required for their tests, which can cost millions of dollars.

The Lab will be initially available to principal investigators (PIs) and research faculty at Carnegie Mellon and open for external use in late March or early April, Jump said. Among the first professors to use the Cloud Lab will be Subha Das, an associate chemistry professor who is using the Emerald Cloud Lab to teach until the Pittsburgh site is online and will be spearheading educational initiatives at the new facility.

Das has developed and taught cloud lab classes since 2020, he said, with the help of the Emerald Cloud Lab. His classes that use the remote laboratory include Welcome to the Future Lab, Science in the Cloud, and Research in the Future Lab — DNA Science in the Cloud. Das is also using the lab in his own research “for automated experiments to study the properties of synthesized and modified DNA,” he said.

Other faculty using the Emerald Cloud Lab who will use the new facility include associate chemistry professor, Oles Isayev, and assistant teaching professor of automated science and computational biology, John Kangas.

“We hope that once the CMU lab is open, there will already be a good library of protocols and experiments for users,” Das said. “It’s quite exciting.”

A particular advantage of using the Cloud Lab is that the online interface configures different lab instruments to be compatible with each other and allows each instrument’s individual coding language to communicate with the system, an innovation by Kleinbaum and Frezza in their pioneering Emerald Cloud Lab. Mechanical arms are automated to do each of the experiments, and human contact is limited to moving the experiments themselves to the next station or to storage.

The Cloud Lab also refines testing by controlling for environmental variables and reducing analysis time. Both of these are important for reproducibility, a critical standard for chemical and biological research. Labs in different locations may struggle to reconstruct the conditions under which an initial test was done, while the Cloud Lab is a stable setting that anyone anywhere can access.

As a safety precaution, the lab is Good Laboratory Practice certified and designated a Biosafety Level 2 facility, which entails a series of safe containment practices intended to prevent hazardous materials from being tested or created and to make the workspace as secure as possible.

In his research, Das said he has appreciated the “good reproducibility in highly repetitive measurements of properties of new types of materials and sequences of DNA with unnatural modifications” that the cloud lab allows. With robotic handling and dispensing of miniscule volumes of solutions, as well as human operators setting up experiments as needed, Das said, the lab will be a powerful innovation in science.

“Most of this research and coding of the cloud experiments is done directly by me,” Das told The Tartan. “I can now queue up experiments from my computer even at home, or while I’m waiting at the gate at an airport.” He said he hopes to engage more trained undergraduate and graduate students in his cloud lab classes to do more groundbreaking research.

Since the Cloud Lab’s instruments take care of many of the finer points of scientific research, the day-to-day operations inside the facility are kept simple. After the initial funding provided by Carnegie Mellon’s Future of Science initiative, the lab will operate solely off income provided by the thread subscription services.

One greater ambition of the Cloud Lab is to motivate the creation of other, similar data aggregation facilities across the country. Building on the precedent of supercomputing labs such as Carnegie Mellon’s and the University of Pittsburgh’s own Pittsburgh Supercomputing Center, these automation facilities could enter into the national infrastructure and streamline the process of scientific research at large. They could be particularly helpful for those not tied to major research institutions who may not be able to afford the equipment required for their investigations.

“Using machine learning and artificial intelligence with the large data sets produced will accelerate our advances,” Meyer said. “It is an amazingly exciting time to be sitting at the forefront of this transformation.”