Chunglab_dz

Deep zoom files for the Kwanghun Chung Lab

This is a utility for creating deep zoom files for the mousebrain website. The inputs are a blockfs volume and an alignment to the reference volume and a layout.json file that coordinates what planes are generated and the outputs are deep zoom directories and XML files.

layout.json

The format of layout.json is a list of dictionaries. Each dictionary defines a plane to be created. The keys to the dictionary are:

• axes: either "xy", "xz" or "yz" to pick either the xy, xz or yz plane for output

• plane: the coordinate perpendicular to the plane defined by axes

• name: the name of the output file within the root directory

• magnification (optional - default=10): the magnification of the reference coordinates, for instance a magnification of 10 gives one pixel every 10th of a coordinate in the reference coordinate frame.

• tile_size (optional - default=254): the size of each image tile

• tile_overlap (optional - default=1): the amount of overlap beetween tiles

• tile_format (optional - default=jpg): the image file format

see layout.json for an example.

chunglab-dz command

The chunglab-dz command creates deep zoom files according its layout file, warping them to the atlas reference using an alignment. It optionally creates a composite color image by combining all of the input channels.

Usage:

chunglab-dz \
  --src <precomputed-blockfs-file> \
  [--src <precomputed-blockfs-file>] \
  [--src <precomputed-blockfs-file] \
  --dest <dest-directory> \
  [--dest <dest-directory>] \
  [--dest <dest-directory>] \
  --alignment <alignment-file> \
  --reference <reference-volume> \
  --layout-file <layout-file> \
  [--n-cores <n-cores>] \
  [--clip <clip>] \
  [--clip <clip>] \
  [--clip <clip>] \
  [--combine] \
  [--skip-if-present]

where

• precomputed-blockfs-file is the path to the precomputed.blockfs file for the volume. This can be specified up to three times for the red, green an blue channels of the combined volume.

• dest-directory is the path to the root directory for deep zoom files. It can be specified up to three times for the three possible channels. Paths in the layout-file are relative to this directory.

• alignment-file is the alignment JSON file that aligns the volume to the reference, e.g. as generated by nuggt-align and rescale-alignment-file.

• reference-volume is the reference volume that the alignment file targets, for instance /nature-protocols-2019-assets/atlas/autofluorescence_25_whole_sagittal.tif.

• layout-file is the JSON layout file (see above for its format)

• n-cores number of processes to use when working in parallel. Defaults to the number of cores on the machine being used.

• clip the intensity at which to clip the image. Defaults to the full range. This should be specified once per channel.

• --combine If present, create a composite image from all channels. This is named by taking the last folder from every channel's destination and concatenating them. For instance if the destinations are /path-to/pv and /path-to/td, the combine destination will be /path-to/pv+td.

• --skip-if-present Don't create a deep zoom image if its .dzi file is present.