AMST2

This is a re-implementation of Alignment to Median Smoothed Template (AMST), a stack alignment for FIB-SEM datasets, originally published here:

• hennies et al. 2020
• original github repo

AMST2 uses the same concept compared to the original AMST workflow, while the implementation is now based on Simple Elastix (Simple ITK) and includes a pre-alignment workflow. To reduce the amount of dependencies as well as to increase robustness of the workflow, the use of SIFT was replaced by Simple Elastix functionality.

Installation

The code was developed in Linux. The snakemake-based workflow can cause issues when working in Windows, we recommend using WSL.

Prerequesite: Install conda on your system, e.g. from https://conda-forge.org/miniforge/

For local execution

mamba create -n amst2-0.3.22-env -c bioconda -c conda-forge --override-channels python=3.11 nibabel napari pyqt opencv zarr=2 vigra pandas snakemake=8
conda activate amst2-0.3.22-env
pip install SimpleITK-SimpleElastix transforms3d ruamel.yaml pyvsnr pytz tzdata
pip install https://github.com/jhennies/squirrel/archive/refs/tags/0.4.8.tar.gz
pip install https://github.com/jhennies/AMST2/archive/refs/tags/0.3.22.tar.gz

For slurm cluster execution

mamba create -n amst2-0.3.22-env -c bioconda -c conda-forge --override-channels python=3.11 nibabel napari pyqt opencv zarr=2 vigra pandas snakemake=8 snakemake-executor-plugin-slurm
conda activate amst2-0.3.22-env
pip install SimpleITK-SimpleElastix transforms3d ruamel.yaml pyvsnr pytz tzdata
pip install https://github.com/jhennies/squirrel/archive/refs/tags/0.4.8.tar.gz
pip install https://github.com/jhennies/AMST2/archive/refs/tags/0.3.22.tar.gz

Usage

To test the installation, we recommend using this dataset: https://www.ebi.ac.uk/empiar/EMPIAR-10311/

I have set up a suitable subset of the EMPIAR-10311 dataset on my owncloud which you can test on in a new directory like so:

wget https://oc.embl.de/index.php/s/RhlFWP3JGnuoIAM/download
unzip download
rm download

Run the test alignment (remove the "--slurm" line if you are running locally):

amst2-wf-nsbs_pre_align-get_default_parameter_file \
  -pi hela-test-dataset/ \
  -po pre-align \
  -p general:resolution:[0.008,0.005,0.005] \
     general:batch_size:8 \
     general:max_cores_per_task:8 \
     nsbs_alignment:batch_size:16 \
     pre_align_to_tif_stack:active:true \
  --slurm \
  --estimate_crop_xy
  
amst2-wf-nsbs_pre_align-run params_pre_align.yaml

Also check out the most recent example for pre alignment and AMST