Microflow: a prototype Nextflow workflow for microscopy image preprocessing

About

This is a prototyping attempt of a common image processing workflow using Nextflow. The steps include:

• Organizing and exporting image metadata into a YAML file.
• Estimating shading profiles and correcting the shading effect.
• Stitching (by M2Stitch) into a Zarr array.

All the steps are written as a Jupyter notebook and run by Papermill in the workflow.

How to use

1. Create a list of files to process (currently AICSImageIO compatible files work).

2. Execute the workflow. The parameters are as follows:

   • -w : The working directory for Nextflow.
   • --input_path_csv : The CSV file for the file list.
   • --common_input_path : The "root" directory of the image files. The output images are stored preserving the subdirectory structure with respect to this directory.
   • --output_path : The directory in which output images are stored.

   nextflow run path/to/microflow \
       -resume \
       -w /path/to/workdir/ \
       --input_path_csv "./file_list.txt" \
       --common_input_path "/mnt/showers/Cell-picker/240119/" \
       --output_path "/path/to/output/"

   One can also specify the following parameters:

   • --shading_estimation_strategy = "timewise" // or "all" : If "timewise", estimate the shading profile for each frame. If "all", use the all frames.
   • --shading_estimation_median_robust = "False" // or "True" : If True, use the intensities within the 2*(Median absolute deviation from median) around the median.
   • --shading_estimation_median_filter_size = 3 : Median filter size for shading correction.
   • --shading_estimation_gaussian_filter_size = 40 : Gaussian filter size for shading correction.
   • --shading_correction_mode = "additive" // or "multiplicative" : The shading correction mode.
   • --shading_correction_local_subtraction_channels = "DIA" : The channel pattern for shading correction by local intensity subtraction.
   • --shading_correction_local_subtraction_scaling = 0.1
   • --shading_correction_local_subtraction_median_disk_size = 4
   • --stitching_stitch_every_t = 0 : The frame skip for stitching. If zero, only use the first frame for stitching.
   • --stitching_target_channel = '10x_Fukai_DIA_IS' : The reference channel for stitching.

Todo

[] Adding other shading correction / stitching methods [] Reorganizing input / output image formats [] Organizing documentation [] Splitting each step to command line tools [] Deploying to nf-core