Parameter Reference

Here we provide parameter options for each workflow in MIAAIM. Commands are added in the

Note

For up-to-date references of parameters for each workflow, visit the Workflow GitHub Repositories

Nextflow

Parameters passed are passed directly using the indicated flags below. See here for a full description of flags that can be passed to Nextflow. We have listed those below that are necessary for MIAAIM to run or are commonly used in our own usage.

Nextflow parameters

Flag	Description	Options
-with-report	generate provenance report
-resume	resume workflow with cached results
--profile	use indicated configuration profile	standard small medium big multi hyper LSFsmall LSFbig_multi
--start-at	start workflow at indicated position	input hdiprep hdireg
--start-at	start workflow at indicated position	hdiprep hdireg
--fixed-image	fixed image for registration	*.ome.tiff *.ome.tif *.tiff *.hdf5 *.h5 *.nii *.imzml
--moving-image	moving image for registration	*.ome.tiff *.ome.tif *.tiff *.hdf5 *.h5 *.nii *.imzml
--fixed-pars	fixed image preprocessing yaml file	*.yaml see HDIprep parameter reference
--moving-pars	moving image preprocessing yaml file	*.yaml see HDIprep parameter reference
--elastix-pars	elastix registration parameters	*.txt see HDIreg parameter reference
--transformix	apply transformation matrix
--transformix-pars	transformix parameters	*.txt see HDIreg parameter reference

Note

When supplying file names as parameters, do not indicate the file’s full path – only include the file name. For example, do not enter --fixed-image path/to/image.ome.tiff. Instead, enter --fixed-image image.ome.tiff.

Warning

If your input file is in the .imzml format, be sure to include the associated .ibd file in the same directory that the image is stored.

Note

You can start a workflow at an intermediate step, such as hdireg, as long as you have the appropriate intermediate files present. hdiprep produces files with the suffix _processed.

Warning

Nextflow does not currently see contents of parameters supplied to the registration workflow. This means that resuming your analysis after changing elastix parameter files will produce cached results. To fix this, change the file name, or start the workflow at the hdireg step without using the resume flag.

HDIprep

Here we list the optional input parameters that can be supplied to the HDIprep workflow through the YAML parameter file. These parameters are passed to both the fixed and moving images with separate YAML files.

–fixed-pars and –moving-pars

YAML Step / Flag	Description	Options
Step : ImportOptions	options for reading image data
flatten:	flatten pixels for dimension reduction	Options: True if compressing images False if histology processing
subsample:	subsample image for compression	Options: True if subsampling pixels False if no subsampling
method:	subsampling method	Options: 'grid' for uniform grid sampling 'random' for random coordinate sampling 'pseudo_random' for random sampling initalized by uniform grids
grid_spacing:	x and y grid spacing for sampling	Options: Example : (5,5)
n:	random / pseudo_random sampling fraction	Options: Ex. : 0.15
masks:	TIFF mask to compress image portion	Options: Ex. : 'moving-mask.tiff'
save_mem:	reduce memory footprint	Options: True for large image compression False if interactive Python code
Step : ProcessingSteps	steps to process images
- RunOptimalUMAP	steady-state compression	Options: n_neighbors nearest neighbors (Ex. 15) landmarks spectral centroids (Ex. 3000) metric UMAP metric (Ex. euclidean) random_state seed (Ex. 1) dim_range range of dimensionalities (Ex. (1,10)) **kwargs kwargs passed to UMAP
- RunUMAP	UMAP compression	**kwargs passed to UMAP
- RunOptimalParametricUMAP	neural network steady state UMAP	Options: n_neighbors nearest neighbors (Ex. 15) landmarks spectral centroids (Ex. 3000) metric UMAP metric (Ex. euclidean) random_state seed (Ex. 1) dim_range range of dimensionalities (Ex. (1,10)) **kwargs kwargs passed to UMAP
- RunParametricUMAP	neural network UMAP compression	**kwargs passed to UMAP
- SpatiallyMapUMAP	reconstruct compressed image
- ApplyManualMask	apply manual mask	Options: mask accessed from ImportOptions
- MedianFilter	median filter (remove salt and pepper noise)	Options: filter_size filter disk size (Ex. 15) parallel use parallel processing (True or False)
- Threshold	create mask by thresholding	Options: type threshold type ('manual' or 'otsu') thresh_value manual threshold value (Ex. 1.0) correction multiply threshold for stringent results (Ex. 1.2)
- Open	morphological closing on mask	Options: disk_size filter disk size (Ex. 15) parallel use parallel processing (True or False)
- Close	resume workflow with cached results	Options: disk_size filter disk size (Ex. 15) parallel use parallel processing (True or False)
- Fill	Fill holes in mask
- ApplyMask	apply mask to image for final processing step
- NonzeroBox	extract image bounding box (for controlled padding)
Step : ExportNifti1	export in the NIfTI format	Options: padding pad to add to images image (Ex. (50,50)) target_size resize image before padding (Ex. (1000,1050))

HDIreg

—elastix-pars

Flag	Description	Options
--p	parameter file(s) for registration	*.txt
--mp	moving image landmark points	*.txt
--fp	fixed image landmark points	*.txt
--fMask	fixed image mask	*.tif

Tip

You can chain together multiple elastix parameter files by supplying multiple inputs. For example, an affine registration followed by a nonlinear one can be implemented as --p affine.txt nonlinear.txt where affine.txt and nonlinear.txt are your parameter files.

—transformix-pars

Flag	Description	Options
--tps	transformation parameter file(s)	*.txt
--target_size	resize image before padding	(Ex. (1000,1050))
--pad	pad to add to images image	(Ex. (50,50))
--trim	number of pixels to trim off edges	(Ex. 50)
--out_ext	aligned image final file format	.ome.tiff .ome.tif .tiff .hdf5 .h5 .nii

Tip

If you use multiple registration parameter files in elastix, then you should add both sets of transformation parameters to the transformix command to receive final results. From the above example, two transformation parameter files would be exported – TransformationParameters.0.txt for the affine registration and TransformationParameters.1.txt for nonlinear. Your transformix call should be --tps TransformationParameters.0.txt TransformationParameters.1.txt