Skip to content

Latest commit

 

History

History
53 lines (40 loc) · 2.47 KB

reconstruction-guide.md

File metadata and controls

53 lines (40 loc) · 2.47 KB
Raw

Automating reconstructions

recOrder uses a configuration-file-based command-line interface (CLI) to perform all reconstructions.

Preparing your data

recOrder is compatible with OME-Zarr, a chunked next generation file format with an open specification. All acquisitions completed with the recOrder plugin will be automatically converted to .zarr format, and existing data can be converted using iohub's convert utility.

Inside a recOrder environment, convert a Micro-Manager TIFF sequence, OME-TIFF, or pycromanager NDTiff dataset with

iohub convert `
    -i ./dataset/ `
    -o ./dataset.zarr

How can I use recOrder's CLI to perform reconstructions?

recOrder's CLI is summarized in the following figure:

The main command reconstruct command is composed of two subcommands: compute-tf and apply-inv-tf.

A reconstruction can be performed with a single reconstruct call. For example:

recorder reconstruct `
    -i ./data.zarr/*/*/* `
    -c ./config.yml `
    -o ./reconstruction.zarr

Equivalently, a reconstruction can be performed with a compute-tf call followed by an apply-inv-tf call. For example:

recorder compute-tf `
    -i ./data.zarr/0/0/0 `
    -c ./config.yml `
    -o ./tf.zarr

recorder apply-inv-tf 
    -i ./data.zarr/*/*/* `
    -t ./tf.zarr `
    -c ./config.yml `
    -o ./reconstruction.zarr

Computing the transfer function is typically the most expensive part of the reconstruction, so saving a transfer function then applying it to many datasets can save time.

Input options

The input -i flag always accepts a list of inputs, either explicitly e.g. -i ./data.zarr/A/1/0 ./data.zarr/A/2/0 or through wildcards -i ./data.zarr/*/*/*. The positions in a high-content screening .zarr store are organized into /row/col/fov folders, so ./input.zarr/*/*/* creates a list of all positions in a dataset.

The recorder compute-tf command accepts a list of inputs, but it only computes the transfer function for the first position in the list. The apply-inv-tf command accepts a list of inputs and applies the same transfer function to all of the inputs, which requires that all positions contain arrays with matching TCZYX shapes.

What types of reconstructions are supported?

See /recOrder/examples/ for a list of example configuration files.

WIP: This documentation will be expanded for each reconstruction type and parameter.