Stream a Zarr from lincbrain.org or dandiarchive.org¶

Before running this code, make sure you have the required dependencies installed. You can install them using pip:

pip install 'dandi>=0.74.3' 'zarr>=3.1.5' matplotlib

Import modules¶

In [18]:

import os
import zarr
import requests
from dandi.dandiapi import DandiAPIClient
from getpass import getpass
from matplotlib import pyplot

# Require zarr v3
if not hasattr(zarr, 'api'):
    raise ImportError(
        f"zarr v3 is required (installed: {zarr.__version__}). "
        "Run: pip install 'zarr>=3.1.5'"
    )
from zarr.api.synchronous import open_group as _zarr_open_group

import os import zarr import requests from dandi.dandiapi import DandiAPIClient from getpass import getpass from matplotlib import pyplot # Require zarr v3 if not hasattr(zarr, 'api'): raise ImportError( f"zarr v3 is required (installed: {zarr.__version__}). " "Run: pip install 'zarr>=3.1.5'" ) from zarr.api.synchronous import open_group as _zarr_open_group

Configuration¶

In [19]:

# API URL (i.e. https://api.lincbrain.org/api or https://api.dandiarchive.org/api)
API_URL = "https://api.lincbrain.org/api"

# Dandiset ID and version
DANDISET_ID = "000056"
DANDISET_VERSION = "draft"

# Path to the OME-Zarr asset within the Dandiset
ASSET_PATH = "derivatives/compressed_camera1/CB1_SH2a_run100_y12_HR.ome.zarr"

# API URL (i.e. https://api.lincbrain.org/api or https://api.dandiarchive.org/api) API_URL = "https://api.lincbrain.org/api" # Dandiset ID and version DANDISET_ID = "000056" DANDISET_VERSION = "draft" # Path to the OME-Zarr asset within the Dandiset ASSET_PATH = "derivatives/compressed_camera1/CB1_SH2a_run100_y12_HR.ome.zarr"

In [20]:

# You will be prompted to enter your DANDI API key.  It can also be set via the DANDI_API_KEY environment variable.
DANDI_API_KEY = os.environ.get("DANDI_API_KEY")
if not DANDI_API_KEY:
    DANDI_API_KEY = getpass("Enter your DANDI API key: ")

# You will be prompted to enter your DANDI API key. It can also be set via the DANDI_API_KEY environment variable. DANDI_API_KEY = os.environ.get("DANDI_API_KEY") if not DANDI_API_KEY: DANDI_API_KEY = getpass("Enter your DANDI API key: ")

Retrieve the asset URL¶

In [21]:

with DandiAPIClient(
    api_url=API_URL,
    token=DANDI_API_KEY,
) as client:
    dandiset = client.get_dandiset(DANDISET_ID, DANDISET_VERSION)
    asset = dandiset.get_asset_by_path(ASSET_PATH)
    zarr_id = asset.zarr

print(f"Dandiset : {dandiset}")
print(f"Asset    : {asset}")
print(f"Zarr ID  : {zarr_id}")

with DandiAPIClient( api_url=API_URL, token=DANDI_API_KEY, ) as client: dandiset = client.get_dandiset(DANDISET_ID, DANDISET_VERSION) asset = dandiset.get_asset_by_path(ASSET_PATH) zarr_id = asset.zarr print(f"Dandiset : {dandiset}") print(f"Asset : {asset}") print(f"Zarr ID : {zarr_id}")

Dandiset : LINC:000056/draft
Asset    : LINC:assets/19c0922e-ce73-4914-9c85-f3d9bcf05cc2
Zarr ID  : e86a8818-a67c-4c1d-b82e-74f589a19a39

Open the Zarr Store¶

In [22]:

if "lincbrain.org" in API_URL:
    # For LINC private data, get the CloudFront signed cookies
    headers = {"Authorization": f"token {DANDI_API_KEY}"}
    session = requests.Session()
    session.get(f"{API_URL}/auth/token", headers=headers)
    response = session.get(f"{API_URL}/permissions/s3/", headers=headers)
    cookies = response.cookies.get_dict()
    cf_url = f"https://neuroglancer.lincbrain.org/zarr/{zarr_id}/"
    storage_options = {"client_kwargs": {"cookies": cookies}}
    zarr_group = _zarr_open_group(cf_url, 
                                  mode='r', 
                                  zarr_format=3, 
                                  storage_options=storage_options)
else:
    # For DANDI public data
    zarr_group = _zarr_open_group(f"s3://dandiarchive/zarr/{zarr_id}/", mode='r')

if "lincbrain.org" in API_URL: # For LINC private data, get the CloudFront signed cookies headers = {"Authorization": f"token {DANDI_API_KEY}"} session = requests.Session() session.get(f"{API_URL}/auth/token", headers=headers) response = session.get(f"{API_URL}/permissions/s3/", headers=headers) cookies = response.cookies.get_dict() cf_url = f"https://neuroglancer.lincbrain.org/zarr/{zarr_id}/" storage_options = {"client_kwargs": {"cookies": cookies}} zarr_group = _zarr_open_group(cf_url, mode='r', zarr_format=3, storage_options=storage_options) else: # For DANDI public data zarr_group = _zarr_open_group(f"s3://dandiarchive/zarr/{zarr_id}/", mode='r')

Read the OME-Zarr metadata¶

In [23]:

print(zarr_group.info)

multiscales = zarr_group.attrs["multiscales"][0]
print(f"Multiscales version : {multiscales['version']}")
print(f"Axes                : {[ax['name'] for ax in multiscales['axes']]}")

for dataset in multiscales["datasets"]:
    path = dataset["path"]
    arr = zarr_group[path]
    print(f"  Level {path} — shape: {arr.shape}, dtype: {arr.dtype}, chunks: {arr.chunks}")

print(zarr_group.info) multiscales = zarr_group.attrs["multiscales"][0] print(f"Multiscales version : {multiscales['version']}") print(f"Axes : {[ax['name'] for ax in multiscales['axes']]}") for dataset in multiscales["datasets"]: path = dataset["path"] arr = zarr_group[path] print(f" Level {path} — shape: {arr.shape}, dtype: {arr.dtype}, chunks: {arr.chunks}")

Name        : 
Type        : Group
Zarr format : 3
Read-only   : True
Store type  : FsspecStore
Multiscales version : 0.4
Axes                : ['z', 'y', 'x']
  Level 0 — shape: (602, 860, 133672), dtype: uint16, chunks: (256, 256, 256)
  Level 1 — shape: (301, 430, 66836), dtype: uint16, chunks: (256, 256, 256)
  Level 2 — shape: (150, 215, 33418), dtype: uint16, chunks: (256, 256, 256)
  Level 3 — shape: (75, 107, 16709), dtype: uint16, chunks: (256, 256, 256)
  Level 4 — shape: (37, 53, 8354), dtype: uint16, chunks: (256, 256, 256)
  Level 5 — shape: (18, 26, 4177), dtype: uint16, chunks: (256, 256, 256)
  Level 6 — shape: (9, 13, 2088), dtype: uint16, chunks: (256, 256, 256)
  Level 7 — shape: (4, 6, 1044), dtype: uint16, chunks: (256, 256, 256)
  Level 8 — shape: (2, 3, 522), dtype: uint16, chunks: (256, 256, 256)
  Level 9 — shape: (1, 1, 261), dtype: uint16, chunks: (256, 256, 256)
  Level 10 — shape: (1, 1, 130), dtype: uint16, chunks: (256, 256, 256)

Create example plot¶

• Plot in the X-Y plane around z=300, y=300, x=68000
• Here is a similar location in Neuroglancer. This link is private to the LINC team, so if it does not load be sure to select Get Neuroglancer Cookies on lincbrain.org.

In [24]:

level0_array = zarr_group['0']
print(f"Level 0 shape: {level0_array.shape}")

z, y, x = 300, 300, 68000
width = 200
slice_xy = level0_array[z, y - width : y + width, x - width : x + width]

fig, ax = pyplot.subplots(1, 1, figsize=(6, 6))
ax.imshow(slice_xy, cmap='gray', aspect='auto')
ax.set_title(f"X-Y plane at Z={z}, centered on Y={y}, X={x}")
ax.set_xlabel("X")
ax.set_ylabel("Y")
pyplot.tight_layout()
pyplot.show()

level0_array = zarr_group['0'] print(f"Level 0 shape: {level0_array.shape}") z, y, x = 300, 300, 68000 width = 200 slice_xy = level0_array[z, y - width : y + width, x - width : x + width] fig, ax = pyplot.subplots(1, 1, figsize=(6, 6)) ax.imshow(slice_xy, cmap='gray', aspect='auto') ax.set_title(f"X-Y plane at Z={z}, centered on Y={y}, X={x}") ax.set_xlabel("X") ax.set_ylabel("Y") pyplot.tight_layout() pyplot.show()

Level 0 shape: (602, 860, 133672)

In [ ]: