This example uses a spatially contiguous, approximately 500 × 500 µm mouse-brain region containing 787 cells. Transcript coordinates and all distance parameters are in micrometers. The full transcript CSV is available from Zenodo.
| Path | Description |
|---|---|
data/Mouse_brain_CosMx_787cells_hyper_result_cb_dedup_1e-05_post_proc.csv |
Final SCRIN colocalization result after directional deduplication and q-value filtering. |
data/Mouse_brain_CosMx_787cells_selected_pair_distances.parquet |
Distance observations for the undirected Apoe-Clu and Gabra2-Gabrb1 pairs, retained within r_dist = 1.0 µm. |
data/Mouse_brain_CosMx_787cells_Gabra2_Gabrb1_cell_events.csv |
Per-cell Gabra2-Gabrb1 colocalization-event counts and median cell coordinates, calculated with r_check = 0.5 µm. |
scripts/plot_distance_distributions.py |
Recreates the selected-pair KDE curves and the two-dimensional random reference. |
scripts/plot_cell_colocalization_map.py |
Recreates the cell-level tissue projection. |
figures/*.png |
Figures displayed below. |
The full transcript CSV and SCRIN intermediate directories are intentionally omitted. They are not required to run the plotting scripts.
Install numpy, pandas, scipy, matplotlib, and a Parquet engine such as pyarrow, then run:
python scripts/plot_distance_distributions.py
python scripts/plot_cell_colocalization_map.pyBoth scripts resolve their default inputs relative to this example directory, so they can be launched from any working directory. Use --help to provide another input or output path. The tissue-projection script flips the x axis by default to match the displayed tissue orientation; pass --no-flip-x to retain the input orientation.
The observed KDE curves show different distance preferences for Apoe-Clu and Gabra2-Gabrb1. The dashed line is the theoretical radial density expected under uniform two-dimensional placement within R = 1.0 µm, f(r) = 2r/R².
Cells are positioned using the median x/y coordinates of all transcripts assigned to each cell. Cells with positive Gabra2-Gabrb1 colocalization-event counts are colored by event count; cells without detected events are shown in light gray.
