Calibration-Free, Nanometer-Scale Single-Molecule Localization with MINFLUX and RASTMIN via Direct Minimum Fitting

Imagine trying to find a single firefly in a vast, dark forest. If you simply scan the area with a flashlight, the firefly's glow might be too faint to pinpoint its exact location, or the beam might be too wide to tell you if it is slightly to the left or right. For decades, scientists have developed ways to see the invisible world of molecules, but to find a single molecule with extreme precision, they often needed a complex setup. They had to first map out the shape of their light beam with incredible care, a process akin to calibrating a ruler before every single measurement. This calibration was necessary because the light used to find these molecules was shaped into a special pattern with a dark spot in the middle, and the molecule would sit right in that darkness. By measuring how much light the molecule gave off when the beam moved around it, scientists could calculate its position. However, this method was slow, required expensive equipment, and was sensitive to even the smallest imperfections in the microscope.

A team of researchers has now shown that this tedious calibration step might not be needed at all. In a new study, they demonstrated that you can find the location of a single molecule with nanometer-scale precision—precision so fine it is measured in billionths of a meter—simply by fitting the raw data to a simple, physical model of the light beam. They tested this idea on two advanced microscopy techniques, one that uses a few specific points of light and another that scans a grid of light across the sample. By skipping the complex calibration and instead using a straightforward mathematical description of the light's shape, they achieved results that were just as accurate as the traditional, calibration-heavy methods, provided enough light was collected. This discovery suggests that the most precise way to see the molecular world could soon become much simpler, cheaper, and more accessible to laboratories around the globe.

The journey to this breakthrough began with a shift in how scientists think about light. Traditional microscopes are limited by the way light bends, making it impossible to distinguish two objects that are closer together than about half the width of a light wave. To overcome this, scientists developed super-resolution techniques that trick the eye or the camera into seeing details smaller than that limit. One such method, known as MINFLUX, works by placing a molecule in the center of a donut-shaped beam of light. The molecule sits in the dark hole of the donut. When the beam moves slightly, the molecule glows more or less depending on how close it is to the edge of the donut. By measuring these changes in brightness, scientists can triangulate the molecule's position with astonishing accuracy. Another method, called RASTMIN, does something similar but scans the donut-shaped beam across a grid of points, creating a small image of the light pattern rather than just a few data points.

For years, the standard way to analyze the data from these experiments was to use a sophisticated statistical tool called maximum likelihood estimation. This tool is incredibly powerful, but it has a major drawback: it requires a perfect, pre-measured map of the light beam. Before every experiment, researchers had to scan a tiny, known bead of light to create a reference image of what the beam looked like. If the microscope drifted, if the lenses were slightly misaligned, or if the light source changed even a little, that reference map would become inaccurate, and the final location of the molecule would be wrong. This made the process fragile and time-consuming, limiting how quickly and easily these techniques could be used.

The researchers behind this new study asked a simple question: if the scanning method creates an image of the light pattern, why not just fit that image to a simple model of what the light should look like? Instead of relying on a pre-measured reference, they proposed using a direct mathematical description of the donut-shaped beam. They tested two different shapes for this model: a simple parabola, which curves like a bowl, and a more complex donut shape that mimics the actual light pattern. They ran thousands of computer simulations, generating fake data that mimicked what a real microscope would see, and then tried to find the molecule's position using both the old calibration-heavy method and their new direct-fitting method.

The results were striking. In the simulations, the new direct-fitting method, especially when using the donut-shaped model, performed almost identically to the gold-standard calibration method. When the researchers collected enough light—specifically, when they detected at least 1,000 photons from the molecule—the difference in precision between the two methods vanished. Both could locate the molecule with an error of less than two nanometers, which is roughly the width of a few atoms. Even more importantly, the new method did not require any prior knowledge of the beam's exact shape. It worked by simply assuming the light followed a physical pattern, and then adjusting that pattern to fit the data it saw.

To ensure this wasn't just a computer trick, the team moved to real-world experiments. They used a custom-built microscope to scan single molecules attached to a DNA structure, a tiny scaffold made of genetic material. They compared the positions found by the traditional calibration method against the positions found by their new direct-fitting method. The results confirmed the simulations: for molecules that emitted 1,000 or more photons, the new method was indistinguishable from the old one in terms of precision. The molecules were located with the same nanometer-scale accuracy.

However, the researchers also discovered a subtle but important nuance. While the simple donut model worked perfectly for finding a single point, it sometimes struggled to preserve the exact shape of a complex structure if the light beam itself was slightly imperfect. Real microscopes often have tiny flaws that make the light donut slightly oval or tilted. When the researchers used a rigid, perfectly symmetrical donut model, the reconstructed image of a square pattern of molecules looked slightly distorted. But when they allowed the model to be flexible—letting the donut stretch or rotate to match the imperfections of the actual light beam—the new method recovered the correct shape perfectly. This showed that the key to success was not just skipping the calibration, but using a model that was flexible enough to adapt to the real world.

The team took their findings a step further by testing the method on a simulated version of a commercial microscope, one that is currently sold to scientists for high-end research. They fed the data from this simulated machine into their new fitting algorithm and compared the results to the manufacturer's own software. The two methods produced images with nearly identical resolution, both resolving details down to about three nanometers. This was a crucial test, as it proved that the new approach could work even with the complex, proprietary data formats used by commercial instruments, not just in idealized academic setups.

The implications of this work are significant for the future of biological imaging. By removing the need for a separate calibration step, the researchers have lowered the barrier to entry for these ultra-precise techniques. A microscope that previously required a specialist to calibrate the light beam every time it was used could now be operated more easily, perhaps even in real-time, allowing scientists to track the movement of molecules as they happen. The method is robust against the small errors and drifts that plague long experiments, and it simplifies the software needed to analyze the data.

This does not mean that every type of advanced microscopy can now skip calibration. The researchers noted that their method works best for scanning techniques where the light pattern is generated by a single optical path. For other variations that use multiple, separate beams of light, a calibration measurement is still strictly necessary to know how those beams relate to each other. But for the growing family of scanning-based techniques like RASTMIN, the path forward is clearer. The study demonstrates that the most precise way to see the molecular world does not always require the most complex tools. Sometimes, a simple, well-chosen model, applied directly to the data, is enough to reveal the truth hidden in the light.