Data analysis

There is a wealth of data analysis methods for the results of single-molecule fluorescence experiments. Here's an overview of the main data- and image-analysis methods used in our lab:

1. Localization and Point Spread Function (PSF) Fitting:
   
   • After acquiring fluorescence data for surface-immobilised fluorescent molecules (in movies containing frames of a field of view of molecules, with typical dimensions of 120 x 50 μm), the first step is to localize the individual molecules. This involves determining the spatial coordinates of the fluorescent signals in each frame of the acquired images.
   • In super-resolution techniques like Single-Molecule Localization Microscopy (SMLM) (e.g., PALM, STORM), we perform precise localization of individual fluorophores. PSF fitting algorithms determine the coordinates of each point source by fitting the observed PSF to a mathematical model.
2. Image reconstruction and Clustering analysis Tracking and Diffusion Analysis:
   
   • Obtaining multiple images of sparce fields of view of molecules, and subsequent high-precision localisation allows used of reconstruction algorithms that combine these images to achieve a higher resolution than the diffraction limit.
   • For single-molecule localisation microscopy data, cluster analysis techniques group localized points into clusters, revealing the spatial distribution of molecular complexes or structures. We typically use density-based clustering (DB-SCAN) or hierarchical clustering, and current explore machine-learning approaches for this task.
3. Image Segmentation:
   
   • Segmentation techniques are used to delineate regions of interest in super-resolution images. This is particularly important for identifying individual bacterial cells, individual cellular structures, or subcellular compartments.
4. Tracking and Diffusion Analysis:
   
   • Tracking algorithms are then applied to link localizations across consecutive frames, allowing the visualization of molecular trajectories (“tracks”).
   • Analyzing the trajectories can report on molecular diffusion. Mean squared displacement (MSD) analysis is commonly used to quantify the displacement of molecules over time and infer their diffusive behavior (e.g., Brownian motion, anomalous diffusion, directed motion).
5. FRET (Fluorescence Resonance Energy Transfer):
   
   • For experiments involving fluorophores that enable FRET measurements, the FRET efficiency can be employed to study molecular interactions and conformational changes. FRET efficiency calculations involve measuring donor and acceptor fluorescence intensities and distances between them. To obtain accurate FRET efficiencies, and calculate the corresponding distances, a series of corrections is needed to ensure high precision and accuracy.
6. Burst Analysis:
   
   • In studies involving molecules diffusing through a confocal spot, burst analysis is used to group fluorescence signals into bursts. This helps in extracting meaningful information about molecular diffusion, structure, and interactions. Such measurements are typically performed using Alternating Laser Excitation to mitigate the effects of fluorophore bleaching, blinking, and incomplete labeling.
7. Correlation Spectroscopy:
   
   • Fluorescence correlation spectroscopy (FCS) and related techniques can be applied to analyze fluctuations in fluorescence signals over time. This enables the measurement of diffusion coefficients, concentration, and molecular interactions in the vicinity of a fluorophore.
8. Hidden Markov Models (HMM):
   
   • HMMs are powerful tools for analysing time-series data in single-molecule experiments. They can be used to identify different states or conformational changes of a molecule based on its fluorescence behaviour.
9. Statistical Analysis:
   
   • Statistical methods, such as bootstrap analysis or Monte Carlo simulations, are often employed to assess the significance of observed trends, estimate uncertainties, and validate results.
10. Machine Learning Approaches:
    
    • Machine learning techniques can be applied for cell segmentation, feature extraction, classification of different molecular states, and even predicting the behaviour of single molecules based on training data.
11. Visualization and Interpretation:
    
    • Visualization tools, such as trajectory plots, scatter plots, and density maps, help in presenting and interpreting the complex data obtained from single-molecule experiments.

These methods offer an advanced understanding of the behaviour and interactions of individual molecules in our biological systems, offering valuable insights into cellular processes and molecular dynamics.