FOCAL3D

FOCAL3D performs density-based clustering of large 3D single-molecule localization microscopy (SMLM) datasets to identify and quantify macromolecular clusters and intracellular structures at nanoscale resolution.

Key Features:

• Density-based clustering: Implements a density-based spatial clustering algorithm specifically tailored for 3D SMLM data.
• Linear scaling performance: Scales linearly with the number of localizations, enabling efficient analysis of large 3D SMLM datasets (unlike traditional algorithms such as DBSCAN, which can be more computationally intensive).
• Objective parameter optimization: Determines an optimal set of clustering parameters through an objective optimization process.
• Robustness to noise and cluster heterogeneity: Demonstrates parametric insensitivity and robust cluster recovery across simulations with varying noise levels and a range of cluster sizes.
• Quantitative accuracy metrics: Uses established metrics such as the F1 score and Silhouette score to evaluate clustering performance.

Scientific Applications:

• 3D dSTORM of nuclear pore complex: Applied to 3D astigmatic dSTORM images of the nuclear pore complex (NPC) in human osteosarcoma cells for clustering of complex heterogeneous structures.
• Quantification of macromolecular organization: Used for quantification and characterization of macromolecular clusters and intracellular structures derived from large SMLM pointillistic datasets.

Methodology:

FOCAL3D applies a density-based spatial clustering algorithm optimized for SMLM, employs objective parameter optimization, scales linearly with localization count, and its performance is validated via simulations and assessed using F1 and Silhouette scores.