Suite2p
Suite2p is a pipeline for processing two-photon calcium imaging data. It handles motion correction, ROI (cell) detection, fluorescence extraction, neuropil subtraction, and spike deconvolution.Input formats
.tiff, .h5, .hdf5, .sbx, .nd2, .raw
Parameters
| Parameter | Range | Default | Description |
|---|---|---|---|
| nplanes | 1–100 | 1 | Number of imaging planes |
| nchannels | 1–4 | 1 | Number of channels |
| Functional channel | 1–4 | 1 | Channel to use for cell detection |
| Frame rate | Float | Auto-detected | Imaging frame rate (Hz) |
| Motion correction | rigid / non-rigid | rigid | Type of motion correction |
| Block size | [Int, Int] | [128, 128] | Block size for non-rigid correction |
| Gaussian smoothing | Float | 0 | Smoothing sigma for cell detection |
| Cell diameter | Integer | 0 (auto) | Expected cell diameter in pixels |
| Detection threshold | 0.1–5 | 1.0 | Scaling factor for detection sensitivity |
| Sparse mode | Toggle | Off | Better for sparse recordings |
| Overlapping ROIs | Toggle | Off | Allow ROIs to overlap |
| Spike deconvolution | Toggle | On | Deconvolve calcium transients to spikes |
| Indicator tau | 0.1–10 | 1.0 | Calcium indicator time constant (seconds) |
| MATLAB export | Toggle | Off | Export results in MATLAB format |
| Combined planes | Toggle | Off | Combine ROIs across planes |
Outputs
| Output | Format | Description |
|---|---|---|
| Cell statistics | NPY | ROI shapes, positions, and properties |
| ROI map | NPY | Spatial footprints of detected cells |
| Raw fluorescence (F) | NPY | Raw fluorescence traces per cell |
| Neuropil (Fneu) | NPY | Neuropil contamination traces |
| Deconvolved spikes | NPY | Estimated spike trains |
| Cell classification | NPY | Classifier labels (cell vs. not-cell) |
| Registered movie | BIN | Motion-corrected movie (optional) |
Presets
| Preset | Description |
|---|---|
| Default | Standard settings for most recordings |
| Fast Processing | Reduced accuracy for quick preview |
| Multi-plane | Configured for multi-plane imaging |
| Dense Labeling | Lower threshold for densely labeled samples |
Compute requirements
| Resource | Requirement |
|---|---|
| GPU | A10G or A100 for large datasets (8–14+ GB VRAM) |
| Duration | ~600 seconds baseline |