A New Way to Count

Ultra-sensitive molecular test gives a more accurate measurement

By Liz Sheeley

When doctors want to know how much of something is in your blood, they use various lab techniques, each designed specifically a particular substance. All of those techniques, although sensitive, take the measurement at the end of the test and only capture the resulting snapshot. With a new method developed by Professor M. Selim Ünlü’s (ECE, MSE, BME) lab, researchers can determine a much more exact measurement by continually observing molecular reactions throughout the test. Their work has been published in Proceedings of the National Academy of Sciences.

Two common molecular analysis tests are ELISA (enzyme-linked immunosorbent assay) and PCR (polymerase chain reaction). The first tests uses antibodies to test for proteins, the second uses synthetic DNA to test for genetic material. They’re commonly used to test for infections, transmittable diseases and show promise in early cancer and traumatic brain injury diagnosis.


Individual molecules bind to an atomically flat surface and detected with Interference Reflectance Imaging Sensing (IRIS). Each molecule is labeled with a gold nanoparticle, which is visible as a faint dark spot when it contacts the surface. IRIS software tracks the how long these binding events last before the bond between the molecule and the surface breaks: each event is circled with in a different color. The actual sensor surface is 50 times larger than this field of view, allowing many simultaneous measurements side-by-side.


In the past decade, advances in these tests have created ways to detect and count very low levels of a target called ultra-sensitive assays. Despite being able to detect single molecules, the read-out isn’t an exact count, but rather a close estimation.

With each test, scientists have designed specific detection molecules that bind to their target. By then measuring the number of bound targets, they can determine the amount of the target in the blood. But sometimes non-target molecules will also bind, which can result in an inaccurate count. This makes it difficult to test for certain classes of important yet diverse molecules, such as micro-RNAs.

“Our new method allows us to step past the trade-off of sensitivity versus specificity,” says Derin Sevenler, a postdoctoral fellow in Ünlü’s lab. “By recording the duration of the individual binding events on video we can better determine which were the target molecule and eliminate the signal from other non-target molecules.”

The work described in the PNAS paper builds on top of Sevenler’s years-long thesis work, which was published in ACS Nano last year. That paper describes a new digital microarray technique that can detect single molecules using gold nanorods; it’s comparable to other state-of-the-art single-molecule detection systems, called digital counting. This new work uses that technique plus an algorithm to process video of the reaction.

The algorithm sorts each binding event and determines if it was the right molecule that bound. And like all other assays, this one must be calibrated. By documenting the number of binding events with solutions where the concentrations of the target molecules are already known, the researchers can build a dataset. Then, when the algorithm determines the number of binding events per hour, it can relate that back to the data and give a readout of the concentration of target molecules.

In the PNAS paper, the researchers added this video-processing algorithm to dynamically track single-binding events, improving the technique from the ACS Nano paper. When the new results were compared to the results from using only the previous method without the video-processing algorithm, they showed that their technique can detect a much smaller concentration—36 times smaller—than it could before.

Traditional endpoint assays like ELISA are not even comparable to this dynamic-tracking technique because they cannot detect single molecules. And dynamic tracking not only outperforms previous digital counting methods because of its 36-times better sensitivity, but it also has a wide field-of-view and can see 12 different reactions at once.

This level of detection could open the door to accurately see micro- and messenger-RNA, and single-strand biomarkers in ultra-low concentrations; those applications could be used for cancer diagnostics, pre-symptomatic viral infection detection and determining how susceptible an infection is to an antibiotic.