transforms
Transforms are applied during the representation stage and you can pass them as a list to the conversion function. For example, this code snippet creates a dataset from RCSB and centers the coordinates of all the proteins:
>>> from proteinshake.datasets import RCSBDataset
>>> from proteinshake.transforms import CenterTransform
>>> da = RCSBDataset().to_voxel(transforms=[CenterTransform()])
To create your own transform simply inherit from proteinshake.transforms.Transform()
and implement the __call__(self, protein)
method.
This function is applied to the protein dictionary of every protein in a dataset.
|
|
A callable object which accepts a protein dictionary and returns an updated version of it. |
|
Do nothing to the protein |
|
Center the coordinates of a protein at atom and residue level. |
|
Apply a random rotation to the coordinate arrays of a protein |
- class Transform[source]
A callable object which accepts a protein dictionary and returns an updated version of it.
- class IdentityTransform[source]
Do nothing to the protein
- class CenterTransform(resolution='residue')[source]
Center the coordinates of a protein at atom and residue level. We use the Ca to compute the center for all the atoms.
- class RandomRotateTransform(resolution='residue', seed=42)[source]
Apply a random rotation to the coordinate arrays of a protein