nipype2pydra 0.2

Basic Usage

To convert a Nipype interface to a Pydra task run the nipype2pydra task command and pass it the conversion specification YAML file and the root of the package to save the generated module, e.g.:

$ nipype2pydra task ants_registration_registration.yaml /path/to/package/root

This will create a module file under the package root directory based on the output_module field in the specification, e.g:

/path/to/package/root/output/module/path

If that is missing and the nipype interface is in the standard nipype.interfaces package, then it will be stored at pydra.tasks. with the same path end.

Conversion Specifications

While the conversions aim to as automatic as possible, there are structural differences between Nipype and Pydra that require manual specification, and it might be desirable to rename some of the inputs/outputs to more intuitive names. Such specifications are stored in YAML format, such as the ANTs registration conversion specification,

task_name: n4_bias_field_correction
nipype_name: N4BiasFieldCorrection
nipype_module: nipype.interfaces.ants.segmentation
inputs:
    omit:
    # list[str] - fields to omit from the Pydra interface
    rename:
    # dict[str, str] - fields to rename in the Pydra interface
    types:
    # dict[str, type] - override inferred types (use "mime-like" string for file-format types,
    # e.g. 'medimage/nifti-gz'). For most fields the type will be correctly inferred
    # from the nipype interface, but you may want to be more specific, particularly
    # for file types, where specifying the format also specifies the file that will be
    # passed to the field in the automatically generated unittests.
        input_image: medimage/nifti1
        mask_image: medimage/nifti1
        weight_image: medimage/nifti1
        bias_image: medimage/nifti1
    metadata:
    # dict[str, dict[str, any]] - additional metadata to set on any of the input fields (e.g. out_file: position: 1)
outputs:
    omit:
    # list[str] - fields to omit from the Pydra interface
    rename:
    # dict[str, str] - fields to rename in the Pydra interface
    types:
    # dict[str, type] - override inferred types (use "mime-like" string for file-format types,
    # e.g. 'medimage/nifti-gz'). For most fields the type will be correctly inferred
    # from the nipype interface, but you may want to be more specific, particularly
    # for file types, where specifying the format also specifies the file that will be
    # passed to the field in the automatically generated unittests.
        output_image: medimage/nifti1
        bias_image: medimage/nifti1
    callables:
    # dict[str, str] - names of methods/callable classes defined in the adjacent `*_callables.py`
    # to set to the `callable` attribute of output fields
    templates:
    # dict[str, str] - `output_file_template` values to be provided to output fields
    requirements:
    # dict[str, list[str]] - input fields that are required to be provided for the output field to be present

Detailed description of the different options to go here

Installation

Nipype2Pydra can be installed for Python >= 3.7 from PyPI with

$ python3 -m pip install nipype2pydra

License

This work is licensed under a Creative Commons Attribution 4.0 International License