STRAND — Subcellular Transcript RNA Architecture and Navigation Database

STRAND is a curated collection of subcellular-resolution spatial transcriptomics datasets, harmonized into a single processed format for downstream analysis (cell-type annotation, RNA localization patterns, colocalization, compartment/RNAflux, segmentation QC).

Overview

• 32 datasets / 77 samples (~191 GB total)
• Platforms: MERFISH, SeqFISH+, Molecular Cartography, STARmap, Xenium, CosMx, MERSCOPE
• Species: human, mouse
• Tissues: brain, liver, gut/intestine, pancreas, and multiple cancer types (ovarian, breast, lung, colorectal, bladder, lymphoma, mesothelioma, ...)

File Organization

Each dataset has its own folder. Datasets with multiple samples/slices contain multiple .pkl files:

Dataset10_Xenium_MouseBrain/
├── Dataset10_1_Xenium_MouseBrain_data_dict.pkl
├── Dataset10_2_Xenium_MouseBrain_data_dict.pkl
└── ...
Dataset26_CosMx_HumanOvarianCancer/
└── Dataset26_CosMx_HumanOvarianCancer_data_dict.pkl

The folder name encodes Dataset<N>_<platform>_<tissue>; the file name additionally carries the sample/slice index.

Data Format

Each *_data_dict.pkl is a Python dict (standard pickle) with the following keys:

Key	Type	Description
data_df	DataFrame	Transcript-level table, one row per detected transcript. Columns: cell, gene, x, y, z, umi (x/y/z in microns, umi = count).
coordinates	DataFrame	Per-cell table: cell centers plus fov and batch columns (each cell's field-of-view / batch membership).
expression	DataFrame	Cell × gene expression matrix (wide format).
cell_boundary	dict	{cell_id: DataFrame(x, y)} — polygon vertices of each cell's segmentation boundary.
nuclear_boundary	dict	{cell_id: DataFrame(x, y)} — polygon vertices of each nucleus (only for datasets with nuclear segmentation / DAPI).
metadata	dict	Dataset-level metadata (platform, tissue, species, etc.).

Notes

• data_df has no fov column — a transcript's FOV is resolved by joining its cell to coordinates (cell → fov).
• CosMx datasets additionally contain an unassigned_data_df (free transcripts with cell = NA).
• Column dtypes may vary across platforms (e.g. CosMx uses category/float32/int16); read tolerantly.

Loading

import pickle

path = "Dataset26_CosMx_HumanOvarianCancer/Dataset26_CosMx_HumanOvarianCancer_data_dict.pkl"
with open(path, "rb") as f:
    d = pickle.load(f)

print(d.keys())
print(d["data_df"].head())       # transcripts: cell, gene, x, y, z, umi
print(d["coordinates"].columns)  # cell centers + fov, batch

---

Source attribution and citation information will be added upon publication.