Imaging data for Figure 5l,m in Boulanger-Weill et al. (2026)

# Imaging data for Figure 5l,m in Boulanger-Weill et al. (2026) This repository contains the processed imaging data underlying Figure 5l,m in Boulanger-Weill et al. (2026). We performed two-photon imaging in 6 larval zebrafish expressing cyto-GCaMP8s. For each fish, 4 imaging planes were recorded following the procedures described in the Methods section of the paper. ## Stimuli The following visual motion stimuli were used: - `constant_100_left`: random dot motion, 100% coherence, leftward- `constant_100_right`: random dot motion, 100% coherence, rightward- `constant_50_left`: random dot motion, 50% coherence, leftward- `constant_50_right`: random dot motion, 50% coherence, rightward- `oscillating_left`: oscillatory random dot motion stimulus whose coherence repeatedly increases and decreases while the net direction remains leftward- `oscillating_right`: oscillatory random dot motion stimulus whose coherence repeatedly increases and decreases while the net direction remains rightward- `switching_100_left`: stimulus initially moves leftward and then suddenly switches to rightward motion- `switching_100_right`: stimulus initially moves rightward and then suddenly switches to leftward motion The exact stimulus input is defined in `dot_motion_coherence_oscillations_switching2.py`. ## Data organization For each fish, the repository contains HDF5 files with segmented fluorescence data. Within each HDF5 file, each imaging plane can be found under: `repeat00_tile000_z<plane>_950nm/preprocessed_data/fish00` Interpolation was performed with `delta t = 0.5 s`. The main contents are: - `cellpose_segmentation`: segmentation-derived data- `imaging_data_channel0_time_averaged`: time-averaged image of the imaging plane- `imaging_information`: imaging metadata at each timestamp (for example laser power/status, PMT gain, lambda half-wave plate orientation)- `stimulus_information`: stimulus timing and identity, with: - column 0: `start_time` (Unix-time stamp) - column 1: `end_time` (Unix-time stamp) - column 2: `stimulus_index` (see stimulus_index_names attribute for naming) ## Contents of `cellpose_segmentation` - `F`: fluorescence of each neuron over time- `masks`: mask image containing all segmented neurons (`0` = background; each neuron has a unique ID)- `stimulus_aligned_dynamics`: fluorescence traces aligned to stimulus presentations - `F` has dimensions: `trials x neurons x interpolated_time` - interpolated time is sampled at 0.5 s intervals- `unit_centroids`: x/y centroid position of each neuron- `unit_contour_masks`: for each cell, a mask in which the cell contour is `True`- `unit_contours`: for each cell, the x/y coordinates of the contour- `unit_masks`: for each cell, a mask in which the full cell area is `True`- `unit_names`: names of all cells (`10000 + cell-ID`) ## TIFF overview files The `.tif` files (for example `fish2026-02-10_10-12-11_fish001_z000.tif`) show the time-averaged imaging planes together with the Cellpose-segmented units. Color code:- gray: not labeled- red: MI cells- blue: MON cells- magenta: SMI cells ## Derived summary file `responses_cell_types.pkl` is a Python pickle file containing the final cell-, trial-, and fish-averaged traces over time for each cell type. These traces are the ones plotted in Figure 5l,m. ## Raw data availability The raw data used to generate these preprocessed data are available upon request. ## Archive format The data are distributed as `.tar.gz` archives. ### Extraction **macOS**- Usually, double-click the `.tar.gz` file- If needed, open Terminal and run: `tar -xzf filename.tar.gz` **Linux**- Open a terminal and run: `tar -xzf filename.tar.gz` **Windows**- We recommend using 7-Zip: - Right-click the file and choose `7-Zip -> Extract Here` - or `7-Zip -> Extract to "filename\"` - On newer Windows systems, you can also use PowerShell or Windows Terminal: `tar -xzf filename.tar.gz` ## Checksum verification Each archive is accompanied by a `.sha256` checksum file for verifying data integrity. **macOS**- Run: `shasum -a 256 -c filename.tar.gz.sha256` **Linux**- Run: `sha256sum -c filename.tar.gz.sha256` **Windows**- In PowerShell, run: `Get-FileHash filename.tar.gz -Algorithm SHA256` Then compare the displayed hash with the hash stored in `filename.tar.gz.sha256`. If the hashes match, the archive is intact. ## Contact If you have any problems extracting or using the files, please contact us.