Analysis & Visualization
CellTracksColab offers advanced capabilities to refine and enhance the quality of your tracking data:
- Track Filtering: Remove short-lived or irrelevant tracks to focus on significant object movements.
- Track Smoothing: Address jitteriness in object tracks using moving average. It is beneficial when the main interest is discerning an object's overall movement direction rather than momentary shifts.
For more details, visit here.
CellTracksColab computes multiple useful track metrics :
- Directionality: Measures the overall direction of a track, from start to finish, in relation to its entire path.
- Tortuosity: Indicates the convolution or twisting of a path. A straight path has a tortuosity of 1, with higher values signifying more winding paths.
- Total Turning Angle: Captures the cumulative angular change as the object moves, providing insights into its overall path deviations.
- Spatial Coverage: Represents the area or volume covered by an object's movement in 2D or 3D space, respectively.
For more details, visit here.
Visualization tools in CellTracksColab facilitate the interpretation of tracking data.
- Users can visualize any track metric of their choice.
- Box plots are generated to portray the distribution of the selected metric in each condition.
- Distinct color-coding for biological repeat aids clarity.
- Statistical insights, such as Cohen's d-value and randomization p-values, are provided, allowing researchers to discern significant patterns in their data.
For more details, visit here.
Quality control ensures the tracking data is consistent, reliable, and outliers-free.
- Hierarchical Clustering: Used to determine the similarity among different fields of view, experimental conditions, and repeats.
- Dendrograms: Visual representations showcasing the hierarchical relationship between data points.
For more details, visit here.
High-dimensional tracking data can be overwhelming, but CellTracksColab offers tools to simplify its exploration:
- UMAP: A powerful dimensionality reduction technique. By converting data into 1D, 2D, or 3D visual representations, UMAP makes complex data structures more interpretable.
- HDBSCAN: This clustering algorithm identifies unique patterns and subpopulations within the data, uncovering hidden behaviors.
- Heatmaps & Track Metrics: By plotting track metrics for each cluster, users can understand the characteristics defining each group. Normalized heatmaps further assist in understanding the variation of metrics across clusters.
- Exemplar Tracks: HDBSCAN can pinpoint representative tracks for each cluster, bridging the gap between abstract data points and tangible examples.
- Fingerprinting: This feature captures the distribution of clusters across different experimental conditions, providing a holistic and unbiased snapshot of the data.
For more details, visit here.
