acryo.pipe package

Module contents

class acryo.pipe.ImageConverter(converter: Callable[[numpy.ndarray[Any, numpy.dtype[numpy.float32]], float], numpy.ndarray[Any, numpy.dtype[numpy.float32]]])

Bases: acryo.pipe._classes._Pipeline

Functionthat convert an image.

compose(other: acryo.pipe._classes.ImageProvider) → acryo.pipe._classes.ImageProvider

compose(other: acryo.pipe._classes.ImageConverter) → acryo.pipe._classes.ImageConverter

Function composition

convert(image: numpy.ndarray[Any, numpy.dtype[numpy.float32]], scale: float) → numpy.ndarray[Any, numpy.dtype[numpy.float32]]

with_scale(scale: float) → Callable[[numpy.ndarray[Any, numpy.dtype[numpy.float32]]], numpy.ndarray[Any, numpy.dtype[numpy.float32]]]

Partialize converter with a given scale.

class acryo.pipe.ImageProvider(provider: Callable[[float], acryo.pipe._classes._R])

Bases: acryo.pipe._classes._Pipeline, Generic[acryo.pipe._classes._R]

Function that provides an image at a given scale.

provide(scale: float) → acryo.pipe._classes._R

Provide an image at a given scale.

acryo.pipe.center_by_mass(*args, **kwargs) → NDArray[np.float32]

Centering an image by its center of mass.

acryo.pipe.closing(*args, **kwargs) → NDArray[np.bool_]

Pipe operation that close (or open) a binary image using a circular structure.

Parameters

radius (float) – Radius of the structure element in nanometer. If negative, opening is applied.

acryo.pipe.converter_function(fn: Callable[[Concatenate[numpy.ndarray, float, _P]], numpy.ndarray]) → Callable[[_P], acryo.pipe._classes.ImageConverter]

Convert a function into a curried function that returns a image converter.

Input function must accept fn(img, scale).

Examples

>>> from scipy import ndimage as ndi
>>> @converter_function
... def gaussian_filter(img, scale, sigma):
...     return ndi.gaussian_filter(img, sigma / scale)
>>> converter = gaussian_filter(1.5)
>>> converter(arr)  # return a Gaussian filtered `arr`

acryo.pipe.dilation(*args, **kwargs) → NDArray[np.bool_]

Pipe operation that dilate (or erode) a binary image using a circular structure.

Parameters

radius (float) – Radius of the structure element in nanometer. If negative, erosion is applied.

acryo.pipe.from_array(*args, **kwargs) → NDArray[np.float32]

An image provider function using existing image array.

This function will provide a subtomogram loader with a resized image from an array. Will be used for the template images or the mask images.

>>> loader.align(
...     template=from_array(img, original_scale=0.28),
...     mask=from_file("path/to/mask.mrc"),
... )

Parameters

• img (np.ndarray) – Input image array. Must be 3D.

• original_scale (float, optional) – If given, this value will be used as the image scale (nm/pixel) instead of the scale extracted from the image metadata.

• tol (float) – Tolerance of the scale difference. If the relative difference is smaller than this, the image will not be resized.

acryo.pipe.from_arrays(*args, **kwargs) → list[NDArray[np.float32]]

acryo.pipe.from_atoms(*args, **kwargs) → NDArray[np.float32]

An image provider function using a point cloud.

Given an array of atoms, such as data extracted from a PDB file, this function can generate a 3D image of the atoms by simply building a histogram.

Parameters

• atoms ((N, 3) array) – Atoms coordinates in nanometer.

• weights (np.ndarray, optional) – weights of the atoms.

• center (tuple of float, optional) – Coordinates of the image center. If not given, the geometric center of the atoms will be used.

acryo.pipe.from_file(*args, **kwargs) → NDArray[np.float32]

An image provider function with rescaling.

This function will provide a subtomogram loader with a resized image from a file. Will be used for the template images or the mask images.

>>> loader.align(
...     template=from_file("path/to/template.mrc"),
...     mask=from_file("path/to/mask.mrc"),
... )

Parameters

• path (path-like) – Path to the image.

• original_scale (float, optional) – If given, this value will be used as the image scale (nm/pixel) instead of the scale extracted from the image metadata.

• tol (float) – Tolerance of the scale difference. If the relative difference is smaller than this, the image will not be resized.

acryo.pipe.from_files(*args, **kwargs) → list[NDArray[np.float32]]

Batch image provider function with rescaling.

This function will provide a subtomogram loader with resized images from files. Will be used for the template images.

>>> from glob import glob
>>> loader.align(
...     template=from_files(glob("path/to/template_*.mrc")),
...     mask=from_file("path/to/mask.mrc"),
... )

Parameters

• paths (iterable of path-like) – Paths to the image.

• original_scale (float, optional) – If given, this value will be used as the image scale (nm/pixel) instead of the scale extracted from the image metadata.

• tol (float) – Tolerance of the scale difference. If the relative difference is smaller than this, the image will not be resized.

acryo.pipe.from_gaussian(*args, **kwargs) → NDArray[np.float32]

An image provider function by a Gaussian function.

This function will provide a Gaussian particle with given shape, sigma and shift from the center.

>>> loader.align(
...     template=from_gaussian(shape=(4.8, 4.8, 4.8), sigma=1.2),
...     mask=from_file("path/to/mask.mrc"),
... )

Parameters

• shape (float or tuple of float) – Shape of the output image in nm.

• sigma (float or tuple of float) – Standard deviation of the Gaussian particle in nm.

• shift (tuple of float, optional) – Shift of the Gaussian particle from the center in nm.

acryo.pipe.gaussian_filter(*args, **kwargs) → NDArray[np.float32]

Gaussian filtering an image.

acryo.pipe.gaussian_smooth(*args, **kwargs) → NDArray[np.float32]

Pipe operation that smooth a binary image using a Gaussian kernel.

Parameters

sigma (float) – Standard deviation of the Gaussian kernel.

acryo.pipe.highpass_filter(*args, **kwargs)

acryo.pipe.lowpass_filter(*args, **kwargs)

acryo.pipe.provider_function(fn: Callable[[Concatenate[float, _P]], acryo.pipe._curry._R]) → Callable[[_P], acryo.pipe._classes.ImageProvider[acryo.pipe._curry._R]]

Convert a function into a curried function that returns a image provider.

Examples

>>> @provider_function
... def provide_random_image(scale, shape):
...     return np.random.random(shape)
>>> provider = provide_random_image((10, 20, 30))
>>> provider(0.18)  # return a (10, 20, 30) array

acryo.pipe.shift(*args, **kwargs) → NDArray[np.float32]

Shift the image by nm.

acryo.pipe.soft_otsu(sigma: float = 1.0, radius: float = 1.0, bins: int = 256)

Pipe operation of soft Otsu thresholding.

This operation binarize an image using Otsu’s method, dilate the edges and smooth the image using a Gaussian kernel.

>>> from acryo.pipe import reader, soft_otsu
>>> loader.align(
...     template=reader("path/to/template.mrc"),
...     mask=soft_otsu(2.0, 2.0),
... )

Parameters

• sigma (float, default is 1.0) – Standard deviation of the Gaussian kernel.

• radius (float, default is 1.0) – Radius of the structure element. If negative, erosion is applied.

• bins (int, default is 256) – Number of bins to build histogram.

acryo.pipe.threshold_otsu(*args, **kwargs) → NDArray[np.bool_]

Pipe operation that binarize an image using Otsu’s method.

Parameters

bins (int, default is 256) – Number of bins to build histogram.