Zarrify

Convert TIFF, TIFF stacks, MRC, N5, and zarr v2 files to OME-Zarr v3.

Install

Requires Pixi.

git clone <repo>
cd zarrify
pixi run dev-install

Run

pixi run zarrify --config config.yaml

Copy a config template from config_examples/ and edit for your data.

Config

src: "/path/to/input"         # tiff, tiff stack dir, .mrc, .n5, or zarr v2 .zarr
dest: "/path/to/output.zarr"

cluster: lsf                  # local | lsf
workers: 100
log_dir: "/path/to/lsf/logs"
extra_directives: ["-P myproject"]

zarr_chunks: [32, 32, 32]     # inner chunk shape
shard_shape: [1024, 1024, 1024]  # outer shard — null to disable
codec: zstd                   # zstd | gzip | blosc
codec_level: 3                # null = per-codec default (zstd:3, gzip:6, blosc:5)

axes: [z, y, x]
scale: [8.0, 8.0, 8.0]
translation: [0.0, 0.0, 0.0]
units: [nanometer, nanometer, nanometer]

optimize_reads: false         # chunk-aligned TIFF reads (single TIFF only)

multiscale: false             # build a multiscale pyramid
ms_workers: 100
data_origin: raw              # raw | labels
antialiasing: false
normalize_voxel_size: false
custom_scale_factors: null    # e.g. [[1,1,1],[2,2,2],[4,4,4]]

CLI flags override config values:

pixi run zarrify --config config.yaml --workers 200 --scale 1.0 1.0 0.058 0.058

LSF example

bsub -n 4 -P myproject -o out.log -e err.log \
  "pixi run zarrify --config config.yaml"

Python API

from zarrify.to_zarr import to_zarr
from zarrify.utils.dask_utils import initialize_dask_client

client = initialize_dask_client("local")
to_zarr("input.mrc", "output.zarr", client)