A Standardized 3D Format for AFM in Structural Biology

In a recent article published in the journal Nature Communications, researchers developed a standardized file format for atomic force microscopy (AFM) data that would harmonize AFM outputs with the optics and visualization frameworks prevalent in structural biology, thereby extending AFM's utility beyond qualitative imaging toward quantitative, integrable structural analysis.

Image Credit: Corona Borealis Studio/Shutterstock.com

Background

AFM data are typically represented as two-dimensional height maps with minimal three-dimensional context, making it difficult to integrate them with volumetric electron density maps or atomic models. In contrast, structural biology formats like MRC/CCP4 (used in cryo-EM) or PDB files (for atomic coordinates) are specifically designed to handle volumetric or atomic-scale data. These formats are optimized for three-dimensional density maps generated by optical, electron, or X-ray techniques.

AFM, however, captures only surface topography and lacks true volumetric depth. As a result, its data are usually stored in standard image formats or proprietary file types, rather than in formats compatible with existing structural biology pipelines. This mismatch has limited AFM’s role in comprehensive structural modeling and validation workflows. Without a standardized way to contextualize AFM data in 3D, researchers miss out on tools for enhanced visualization, dynamic interpretation, and cross-validation with other structural modalities.

The Current Study

The key innovation lies in an algorithmic pipeline that converts AFM surface height data into a format comparable to the volumetric density maps used in cryo-EM. The process begins with localization AFM (LAFM), an advanced technique that captures high-resolution, localization-based surface data. From there, automated peak detection and clustering algorithms identify key surface features, such as peaks and valleys, within the AFM images.

These identified features are then computationally integrated into a three-dimensional probability density function. At the heart of the method is a mapping process that projects the height data onto a 3D grid. The original optical signals, generated by the laser-cantilever interaction, are transformed into a density profile using Gaussian mixture models. This approach reconstructs the spatial information embedded in the raw signals into a volumetric representation.

The result is a density map that preserves the precise geometry of the AFM surface features while presenting the data in a format compatible with standard optical visualization tools and computational modeling platforms.

This 3D density map, stored as a file with an extension '.afm', mimics the structure of the MRC format used in cryo-EM, with the key advantage of being directly compatible with existing visualization and analysis software such as Chimera. The authors further develop a force field generation method called Molecular Dynamics Flexible Fitting (MDFF), which employs the optical-like density maps as physical biases in molecular dynamics simulations. The underlying principle relies on the fact that gradients within the density map provide optical-like force vectors that can steer atomic models into conformations consistent with observed surface features.

Results and Discussion

The methodology successfully produces high-resolution 3D density maps that are fully compatible with standard tools used in structural biology. Using experimental AFM data, the authors showcase how surface topographies of membrane proteins and other biomolecules can be converted into these volumetric formats. The resulting maps retain key optical features captured by AFM, such as protrusions, depressions, and conformational shifts, making their correspondence to physical structures visually intuitive.

A key strength of this approach lies in its preservation of the optical characteristics inherent to AFM data. Since the density maps are derived from laser deflection signals, they retain the original optical signatures tied to the surface features. This not only ensures geometric accuracy but also provides a bridge between the raw optical measurements and computational structural analysis.

Furthermore, the researchers show that these AFM-derived density maps are effective as force fields in MDFF simulations, enabling the physical modeling of biomolecular structures from surface topographies. The use of the optical-like density enables the application of realistic biases, steering atomic models to conform with actual surface features observed during live, near-native experiments. These simulations reveal conformational states and transition pathways that are difficult to capture by other structural techniques. The models derived from this technique could be validated against existing high-resolution structures or further refined with experimental data from other modalities.

From an optics standpoint, the paper highlights how this transformation builds on the photon-mechanical detection pathway central to AFM. The laser deflections, typically used to infer surface topography, are reinterpreted as volumetric probability densities, enabling their visualization and manipulation within the 3D frameworks commonly used in structural biology software.

This reinterpretation emphasizes the critical role of AFM’s optical detection mechanism in generating a physically meaningful density representation. Rather than limiting the data to flat surface maps, the method translates optical signals into rich, volumetric datasets that retain depth and detail. The result is a more nuanced and spatially expressive visualization of surface features, grounded in the optical measurements themselves.

Conclusion

In practical terms, this methodology allows researchers to utilize AFM data for detailed structural analysis, dynamic studies, and hypothesis-driven modeling in a manner compatible with tools they are already familiar with. It enhances the potential for cross-validation with other structural techniques, fosters the integration of AFM into comprehensive structural workflows, and promotes broader adoption of AFM data in the elucidation of biomolecular mechanisms. Ultimately, this work underscores how the optical detection principle underpinning AFM can be exploited to advance structural biology, transforming a surface imaging technique into a powerful component of the integrative structural toolkit.