Algorithm rapidly identifies cell types to match patients with effective cancer therapies

Researchers at VCU Massey Comprehensive Cancer Center have developed a novel algorithm that could provide a revolutionary tool for determining the best options for patients—both in the treatment of cancer and in the prescription of medicines.

As recently published in Nature Communications, Jinze Liu, Ph.D., and Kevin Byrd, D.D.S., Ph.D., created Threshold-based Assignment of Cell Types from Multiplexed Imaging Data (TACIT), which assigns cell identities based on cell-marker expression profiles. TACIT cuts down cell identification time from over a month to just minutes—saving researchers valuable time and resources.

TACIT—developed by Liu, a research member at Massey and a professor in the Department of Biostatistics at the VCU School of Public Health, and Byrd, an associate research member at Massey and assistant professor of oral and craniofacial molecular biology at the VCU School of Dentistry—uses data from over 5 million cells across major body systems like the brain, gut, and oral glands to distinguish cells, providing superior accuracy and scalability when compared with existing models, which oftentimes lack the power to separate expected cell populations due to limited marker sets.

“We’re using artificial intelligence to increase efficiency and also the accuracy of diagnosis,” Liu said. “And as we gain more data, TACIT’s ability to increase positive patient outcomes will only multiply.”

In their publication, Liu and Byrd demonstrated how TACIT outperformed three existing unsupervised methods in accuracy and scalability while also integrating cell types and states to reveal new cellular associations. TACIT showed strong agreement between different kinds of tests—like genetic and protein data—making results more trustworthy. For patients, this could mean getting a diagnosis sooner, avoiding unnecessary treatments, or being matched with a clinical trial that’s more likely to help. And for doctors, it provides a powerful way to see what’s really going on inside the body.

The beneficial applications of TACIT are wide-ranging.

“One of our goals as scientists is to identify good spatial biomarkers for clinical trials so that we can predict patient responses to the trial before they even are enrolled,” said Liu. “We have already been working with multiple principal investigators on [VCU’s] campus to include spatial biology into clinical trials, and TACIT can provide that guidance so that we can make sure the clinical trial patients are receiving the best possible treatments.”

Byrd added, “You could use TACIT to get the right patient into the trial—and as importantly—not put the wrong patient in the trial. Right now, we don’t have a very good tool for that, but this is quite powerful to do it.”

Liu and Byrd also see benefits of TACIT in the pharmacological setting, where the algorithm can utilize RNA markers to help drive care.

“If you tell a patient they can’t be part of a clinical trial, that’s not great news, especially if nothing else is offered,” said Byrd. “But these RNA markers are actually quite good and scalable, which allows us to predict drugs and outcomes that may be useful for patients.

“We have a large repository of [FDA] approved drugs we can map onto the tissue samples. Imagine if you could tell a patient, ‘Here’s an already [FDA] approved drug.’ So that patient doesn’t get recruited as part of a trial for a new investigational drug that they might not need or might not see as much benefit from.”

Additionally, TACIT works across multiple spatial biology applications, allowing Liu and Byrd to build upon the existing datasets to further enhance the algorithm’s work. “We sometimes joke that TACIT is like a Rosetta Stone,” Byrd said. “You can see how all these different data types all become the same language, and for us, we can build upon that. There is a huge opportunity to use TACIT in a lot of ways, from proteins to organ systems to different disease types.”

Liu and Byrd’s work also demonstrates a novel technology in which they captured both slide proteomics and transfer proteomics, finding ways to link the two different pieces of equipment in order to create cell multi-omics, allowing researchers to study multiple markers at once. Previously, researchers could only utilize single-cell omics.