Interactive Inspection
The inspect command provides a graphical user interface (GUI) to interactively explore and visualize your clustering results. This is particularly useful for validating the quality of clusters and examining individual traces within them.
Usage
To start the interactive inspection tool, navigate to your project directory and run:
icluto inspect <output_folder>
<output_folder>: The path to the folder containing the results of your clustering run. This folder should containgood_traces.npy(or.bin) and one or more subdirectories withlabels.npyfiles.
Features
The inspection tool opens a window with three main panels:
- Cluster Density (Left): A 2D histogram showing the distribution of all traces in the currently selected cluster.
- Trace View (Middle): A plot of a single trace from the current cluster.
- 1D Histogram (Right): A comparison of the conductance distribution of the current trace (red) against the entire cluster (gray).
Controls
Navigation
- Run Controls: Use
< RunandRun >to switch between different clustering methods or parameter sets found in the output folder. - Cluster Controls: Use
< ClustandClust >to cycle through the unique clusters within the current run. - Trace Controls: Use
< TrandTr >to navigate through individual traces inside the selected cluster.
Keyboard Shortcuts
[!NOTE] Currently, the tool primarily uses the graphical buttons for interaction.
Data Requirements
For the inspect command to work correctly, your output folder must follow this structure:
output_folder/
├── good_traces.npy (or good_traces.bin)
└── run_name/
└── labels.npy
The tool automatically scans for all labels.npy files and allows you to switch between them seamlessly.