New Approach to Counting Molecules in Fluorescence Microscopy

Researchers from HHMI Janelia have developed a method called blinx to estimate the number of individual molecules in a single spot of light detected by a fluorescence microscope. The challenge is similar to trying to count the number of frogs at a pond by listening to the overall volume of croaks, rather than distinguishing each frog’s call. In fluorescence microscopy, the spot of light represents the combined brightness of many fluorescently labeled molecules, and the intensity of this spot changes over time as molecules blink on and off. The resolution of the microscope limits the ability to observe individual molecules directly, so only the total intensity is measured.

To address this, the team modeled the entire path of light through the microscope system, creating a “trace” that reflects the spot’s intensity over time and depends on various system parameters. Using machine learning, they fit this model to actual intensity traces from microscope images, adjusting parameters to best match the observed data. This approach allows them to infer unknown quantities, such as the number of molecules present in a single spot.

Unlike previous methods that provide a single count, blinx generates a probability distribution of possible molecule numbers, giving researchers a sense of confidence in the results and indicating when the data may be insufficient for a definitive answer. According to Jan Funke, senior author of the study published in Nano Letters, the model can communicate uncertainty by showing when it cannot confidently determine a count.

“Sometimes the data just doesn’t support a single answer. There might be so much fluctuation that the information is just not there,” explained Funke. “This model has the capability to tell you: I really don’t know.”

The method is also capable of counting more molecules per spot than earlier techniques, which could be valuable for identifying proteins based on their amino acid composition.