Creating better at-home tests for viruses

Chemistry researchers at the University of Cincinnati are developing better biosensors they hope will replace diagnostics such as the nearly 1 billion COVID tests distributed to homes in the United States during the pandemic.

UC College of Arts and Sciences Assistant Professor Pietro Strobbia is developing new at-home diagnostics for infectious diseases.

"During COVID we had some issues with home diagnostics. Home tests aren't sensitive enough to accurately detect many infectious diseases," he said.

Biosensors are devices used in the rapid and sensitive detection of substances at the molecular level. Existing diagnostics such as polymerase chain reaction tests are more accurate than the antigen tests most people used during the pandemic. But they also are far more expensive, he said.

"We learned a lot from COVID. But we want to be ready for the next COVID, the next dangerous virus," he said.

To that end, Strobbia and his research partners, UC Professors George Stan and Ruxandra Dima, received a $327,000 grant from the National Institutes of Health to develop new surface-enhanced Raman scattering sensors using artificial intelligence.

"What we want to do is build a library of sensors that can detect many targets and use AI to read the sequences and understand which sequences will work best for certain targets," Strobbia said.

The long-term goal of this project is to create a sensing platform capable of detecting multiple genetic biomarkers in an at-home test. Surface-enhanced Raman scattering is already used in lots of industries to detect contaminants in food or in environmental analyses.

These sensors not only can detect the presence of virus but the amount of virus present.

As part of the project, Strobbia's lab invested in a new robot that can prepare the project's many hundreds of fluidic sensors, each of which contains a precise mix of chemicals.

UC doctoral student Steven Quarin said using pipettes to prepare the sensors by hand is tedious and time-consuming.

"I know how long it takes me to prepare a single row in a 96-well plate," he said. "When you're pipetting, it gets very repetitive. It's monotonous work. I'm confident I'm not making mistakes, but there are times when I ask, 'Did I do this right?'"

Strobbia said the robot is reliably accurate and can prepare sensors with less risk of contamination.

Strobbia's research partners, computational chemists Dima and Stan, said they were excited to join the project to develop efficient and cost-effective biosensors.

Stan said biosensor development is hindered by our limited understanding of the optimal biophysical and biochemical properties required for functional DNA sequences.

"Our collaboration aims to develop an AI-based approach that autonomously generates functional DNA sequences optimized for the viral target," he said. "This machine learning approach uses a large library of DNA sequences to learn the 'language' based on the DNA alphabet of four nucleotides."

Stan said since artificial intelligence gets more accurate when it has access to larger data sets, a big part of the project will be to provide a large library of functional DNA sequences.

"In turn, this large DNA library will be used to train the AI-based model," Stan said. "Thus, the availability of this automated pipeline in the Department of Chemistry opens up exciting avenues for biosensor design that combine state-of-the-art experimental and AI-based tools."

Provided by University of Cincinnati