pytme 0.2.0__cp311-cp311-macosx_14_0_arm64.whl → 0.2.1__cp311-cp311-macosx_14_0_arm64.whl

tme/orientations.py

@@ -1,5 +1,5 @@
 #!python3
-""" Handle template matching peaks and convert between formats.
+""" Handle template matching orientations and conversion between formats.
 
     Copyright (c) 2024 European Molecular Biology Laboratory
 
@@ -8,6 +8,7 @@
 import re
 from collections import deque
 from dataclasses import dataclass
+from string import ascii_lowercase
 from typing import List, Tuple, Dict
 
 import numpy as np
@@ -17,7 +18,48 @@ from scipy.spatial.transform import Rotation
 @dataclass
 class Orientations:
     """
-    Handle template matching peaks and convert between formats.
+    Handle template matching orientations and conversion between formats.
+
+    Examples
+    --------
+    The following achieves the minimal definition of an :py:class:`Orientations` instance
+
+    >>> import numpy as np
+    >>> from tme import Orientations
+    >>> translations = np.random.randint(low = 0, high = 100, size = (100,3))
+    >>> rotations = np.random.rand(100, 3)
+    >>> scores = np.random.rand(100)
+    >>> details = np.full((100,), fill_value = -1)
+    >>> orientations = Orientations(
+    >>>     translations = translations,
+    >>>     rotations = rotations,
+    >>>     scores = scores,
+    >>>     details = details,
+    >>> )
+
+    The created ``orientations`` object can be written to disk in a range of formats.
+    See :py:meth:`Orientations.to_file` for available formats. The following creates
+    a STAR file
+
+    >>> orientations.to_file("test.star")
+
+    :py:meth:`Orientations.from_file` can create :py:class:`Orientations` instances
+    from a range of formats, to enable conversion between formats
+
+    >>> orientations_star = Orientations.from_file("test.star")
+    >>> np.all(orientations.translations == orientations_star.translations)
+    True
+
+    Parameters
+    ----------
+    translations: np.ndarray
+        Array with translations of each orientations (n, d).
+    rotations: np.ndarray
+        Array with euler angles of each orientation in zxy convention (n, d).
+    scores: np.ndarray
+        Array with the score of each orientation (n, ).
+    details: np.ndarray
+        Array with additional orientation details (n, ).
     """
 
     #: Return a numpy array with translations of each orientation (n x d).
@@ -32,6 +74,29 @@ class Orientations:
     #: Return a numpy array with additional orientation details (n, ).
     details: np.ndarray
 
+    def __post_init__(self):
+        self.translations = np.array(self.translations).astype(np.float32)
+        self.rotations = np.array(self.rotations).astype(np.float32)
+        self.scores = np.array(self.scores).astype(np.float32)
+        self.details = np.array(self.details).astype(np.float32)
+        n_orientations = set(
+            [
+                self.translations.shape[0],
+                self.rotations.shape[0],
+                self.scores.shape[0],
+                self.details.shape[0],
+            ]
+        )
+        if len(n_orientations) != 1:
+            raise ValueError(
+                "The first dimension of all parameters needs to be of equal length."
+            )
+        if self.translations.ndim != 2:
+            raise ValueError("Expected two dimensional translations parameter.")
+
+        if self.rotations.ndim != 2:
+            raise ValueError("Expected two dimensional rotations parameter.")
+
     def __iter__(self) -> Tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray]:
         """
         Iterate over the current class instance. Each iteration returns a orientation
@@ -77,7 +142,17 @@ class Orientations:
         filename : str
             The name of the file where the orientations will be saved.
         file_format : type, optional
-            The format in which to save the orientations. Supported formats are 'text' and 'relion'.
+            The format in which to save the orientations. Defaults to None and infers
+            the file_format from the typical extension. Supported formats are
+
+            +---------------+----------------------------------------------------+
+            | text          | pyTME's standard tab-separated orientations file   |
+            +---------------+----------------------------------------------------+
+            | relion        | Creates a STAR file of orientations                |
+            +---------------+----------------------------------------------------+
+            | dynamo        | Creates a dynamo table                             |
+            +---------------+----------------------------------------------------+
+
         **kwargs : dict
             Additional keyword arguments specific to the file format.
 
@@ -120,8 +195,14 @@ class Orientations:
         The file is saved with a header specifying each column: z, y, x, euler_z,
         euler_y, euler_x, score, detail. Each row in the file corresponds to an orientation.
         """
+        naming = ascii_lowercase[::-1]
         header = "\t".join(
-            ["z", "y", "x", "euler_z", "euler_y", "euler_x", "score", "detail"]
+            [
+                *list(naming[: self.translations.shape[1]]),
+                *[f"euler_{x}" for x in naming[: self.rotations.shape[1]]],
+                "score",
+                "detail",
+            ]
         )
         with open(filename, mode="w", encoding="utf-8") as ofile:
             _ = ofile.write(f"{header}\n")
@@ -158,9 +239,6 @@ class Orientations:
         References
         ----------
         .. [1]  https://wiki.dynamo.biozentrum.unibas.ch/w/index.php/Table
-
-        The file is saved with a standard header used in Dynamo STAR files.
-        Each row in the file corresponds to an orientation.
         """
         with open(filename, mode="w", encoding="utf-8") as ofile:
             for index, (translation, rotation, score, detail) in enumerate(self):
@@ -316,8 +394,18 @@ class Orientations:
         filename : str
             The name of the file from which to read the orientations.
         file_format : type, optional
-            The format of the file. Currently, only 'text' format is supported.
-        **kwargs : dict
+            The format of the file. Defaults to None and infers
+            the file_format from the typical extension. Supported formats are
+
+            +---------------+----------------------------------------------------+
+            | text          | pyTME's standard tab-separated orientations file   |
+            +---------------+----------------------------------------------------+
+            | relion        | Creates a STAR file of orientations                |
+            +---------------+----------------------------------------------------+
+            | dynamo        | Creates a dynamo table                             |
+            +---------------+----------------------------------------------------+
+
+        **kwargs
             Additional keyword arguments specific to the file format.
 
         Returns
@@ -330,11 +418,18 @@ class Orientations:
         ValueError
             If an unsupported file format is specified.
         """
-        mapping = {"text": cls._from_text, "relion": cls._from_relion_star}
+        mapping = {
+            "text": cls._from_text,
+            "relion": cls._from_relion_star,
+            "tbl": cls._from_tbl,
+        }
         if file_format is None:
             file_format = "text"
+
             if filename.lower().endswith(".star"):
                 file_format = "relion"
+            elif filename.lower().endswith(".tbl"):
+                file_format = "tbl"
 
         func = mapping.get(file_format, None)
         if func is None:
@@ -375,25 +470,28 @@ class Orientations:
         """
         with open(filename, mode="r", encoding="utf-8") as infile:
             data = [x.strip().split("\t") for x in infile.read().split("\n")]
-            _ = data.pop(0)
 
+        header = data.pop(0)
         translation, rotation, score, detail = [], [], [], []
         for candidate in data:
             if len(candidate) <= 1:
                 continue
-            if len(candidate) != 8:
-                candidate.append(-1)
 
-            candidate = [float(x) for x in candidate]
-            translation.append((candidate[0], candidate[1], candidate[2]))
-            rotation.append((candidate[3], candidate[4], candidate[5]))
-            score.append(candidate[6])
-            detail.append(candidate[7])
+            translation.append(
+                tuple(
+                    candidate[i] for i, x in enumerate(header) if x in ascii_lowercase
+                )
+            )
+            rotation.append(
+                tuple(candidate[i] for i, x in enumerate(header) if "euler" in x)
+            )
+            score.append(candidate[-2])
+            detail.append(candidate[-1])
 
-        translation = np.vstack(translation).astype(int)
-        rotation = np.vstack(rotation).astype(float)
-        score = np.array(score).astype(float)
-        detail = np.array(detail).astype(float)
+        translation = np.vstack(translation)
+        rotation = np.vstack(rotation)
+        score = np.array(score)
+        detail = np.array(detail)
 
         return translation, rotation, score, detail
 
@@ -448,20 +546,15 @@ class Orientations:
         ret = cls._parse_star(filename=filename, delimiter=delimiter)
         ret = ret["data_particles"]
 
-        translation = (
-            np.vstack(
-                (ret["_rlnCoordinateZ"], ret["_rlnCoordinateY"], ret["_rlnCoordinateX"])
-            )
-            .astype(np.float32)
-            .astype(int)
-            .T
+        translation = np.vstack(
+            (ret["_rlnCoordinateZ"], ret["_rlnCoordinateY"], ret["_rlnCoordinateX"])
         )
+        translation = translation.astype(np.float32).T
 
-        rotation = (
-            np.vstack((ret["_rlnAngleRot"], ret["_rlnAngleTilt"], ret["_rlnAnglePsi"]))
-            .astype(np.float32)
-            .T
+        rotation = np.vstack(
+            (ret["_rlnAngleRot"], ret["_rlnAngleTilt"], ret["_rlnAnglePsi"])
         )
+        rotation = rotation.astype(np.float32).T
 
         rotation = Rotation.from_euler("xyx", rotation, degrees=True)
         rotation = rotation.as_euler(seq="zyx", degrees=True)
@@ -470,6 +563,33 @@ class Orientations:
 
         return translation, rotation, score, detail
 
+    @staticmethod
+    def _from_tbl(
+        filename: str, **kwargs
+    ) -> Tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray]:
+        with open(filename, mode="r", encoding="utf-8") as infile:
+            data = infile.read().split("\n")
+        data = [x.strip().split(" ") for x in data if len(x.strip())]
+
+        if len(data[0]) != 38:
+            raise ValueError(
+                "Expected tbl file to have 38 columns generated by _to_tbl."
+            )
+
+        translations, rotations, scores, details = [], [], [], []
+        for peak in data:
+            rotation = Rotation.from_euler(
+                "xyx", (peak[6], peak[7], peak[8]), degrees=True
+            )
+            rotations.append(rotation.as_euler(seq="zyx", degrees=True))
+            scores.append(peak[9])
+            details.append(-1)
+            translations.append((peak[25], peak[24], peak[23]))
+
+        translations, rotations = np.array(translations), np.array(rotations)
+        scores, details = np.array(scores), np.array(details)
+        return translations, rotations, scores, details
+
     def get_extraction_slices(
         self,
         target_shape: Tuple[int],
@@ -504,55 +624,55 @@ class Orientations:
         SystemExit
             If no peak remains after filtering, indicating an error.
         """
-        left_pad = np.divide(extraction_shape, 2).astype(int)
-        right_pad = np.add(left_pad, np.mod(extraction_shape, 2)).astype(int)
+        right_pad = np.divide(extraction_shape, 2).astype(int)
+        left_pad = np.add(right_pad, np.mod(extraction_shape, 2)).astype(int)
+
+        peaks = self.translations.astype(int)
+        obs_beg = np.subtract(peaks, left_pad)
+        obs_end = np.add(peaks, right_pad)
 
-        obs_start = np.subtract(self.translations, left_pad)
-        obs_stop = np.add(self.translations, right_pad)
+        obs_beg = np.maximum(obs_beg, 0)
+        obs_end = np.minimum(obs_end, target_shape)
 
-        cand_start = np.subtract(np.maximum(obs_start, 0), obs_start)
-        cand_stop = np.subtract(obs_stop, np.minimum(obs_stop, target_shape))
-        cand_stop = np.subtract(extraction_shape, cand_stop)
-        obs_start = np.maximum(obs_start, 0)
-        obs_stop = np.minimum(obs_stop, target_shape)
+        cand_beg = left_pad - np.subtract(peaks, obs_beg)
+        cand_end = left_pad + np.subtract(obs_end, peaks)
 
         subset = self
         if drop_out_of_box:
-            stops = np.subtract(cand_stop, extraction_shape)
+            stops = np.subtract(cand_end, extraction_shape)
             keep_peaks = (
                 np.sum(
-                    np.multiply(cand_start == 0, stops == 0),
+                    np.multiply(cand_beg == 0, stops == 0),
                     axis=1,
                 )
-                == self.translations.shape[1]
+                == peaks.shape[1]
             )
             n_remaining = keep_peaks.sum()
             if n_remaining == 0:
                 print(
                     "No peak remaining after filtering. Started with"
-                    f" {self.translations.shape[0]} filtered to {n_remaining}."
+                    f" {peaks.shape[0]} filtered to {n_remaining}."
                     " Consider reducing min_distance, increase num_peaks or use"
                     " a different peak caller."
                 )
-                exit(-1)
 
-            cand_start = cand_start[keep_peaks,]
-            cand_stop = cand_stop[keep_peaks,]
-            obs_start = obs_start[keep_peaks,]
-            obs_stop = obs_stop[keep_peaks,]
+            cand_beg = cand_beg[keep_peaks,]
+            cand_end = cand_end[keep_peaks,]
+            obs_beg = obs_beg[keep_peaks,]
+            obs_end = obs_end[keep_peaks,]
             subset = self[keep_peaks]
 
-        cand_start, cand_stop = cand_start.astype(int), cand_stop.astype(int)
-        obs_start, obs_stop = obs_start.astype(int), obs_stop.astype(int)
+        cand_beg, cand_end = cand_beg.astype(int), cand_end.astype(int)
+        obs_beg, obs_end = obs_beg.astype(int), obs_end.astype(int)
 
         candidate_slices = [
             tuple(slice(s, e) for s, e in zip(start_row, stop_row))
-            for start_row, stop_row in zip(cand_start, cand_stop)
+            for start_row, stop_row in zip(cand_beg, cand_end)
         ]
 
         observation_slices = [
             tuple(slice(s, e) for s, e in zip(start_row, stop_row))
-            for start_row, stop_row in zip(obs_start, obs_stop)
+            for start_row, stop_row in zip(obs_beg, obs_end)
         ]
 
         if return_orientations:

tme/preprocessing/_utils.py

@@ -133,11 +133,11 @@ def fftfreqn(
     NDArray
         The sample frequencies.
     """
-    center = backend.astype(backend.divide(shape, 2), backend._default_dtype_int)
+    center = backend.astype(backend.divide(shape, 2), backend._int_dtype)
 
-    norm = np.ones(3)
+    norm = np.ones(len(shape))
     if sampling_rate is not None:
-        norm = backend.multiply(shape, sampling_rate).astype(int)
+        norm = backend.astype(backend.multiply(shape, sampling_rate), int)
 
     if shape_is_real_fourier:
         center[-1] = 0
@@ -151,9 +151,9 @@ def fftfreqn(
     indices = backend.transpose(indices)
 
     if compute_euclidean_norm:
-        backend.square(indices, indices)
+        indices = backend.square(indices)
         indices = backend.sum(indices, axis=0)
-        indices = backend.sqrt(indices)
+        backend.sqrt(indices, out=indices)
 
     return indices
 
@@ -174,3 +174,15 @@ def crop_real_fourier(data: NDArray) -> NDArray:
     """
     stop = 1 + (data.shape[-1] // 2)
     return data[..., :stop]
+
+
+def shift_fourier(data: NDArray, shape_is_real_fourier: bool = False):
+    shift = backend.add(
+        backend.astype(backend.divide(data.shape, 2), int),
+        backend.mod(data.shape, 2),
+    )
+    if shape_is_real_fourier:
+        shift[-1] = 0
+
+    data = backend.roll(data, shift, tuple(i for i in range(len(shift))))
+    return data

tme/preprocessing/composable_filter.py

@@ -8,6 +8,7 @@
 from typing import Dict
 from abc import ABC, abstractmethod
 
+
 class ComposableFilter(ABC):
     """
     Strategy class for composable filters.
@@ -27,4 +28,4 @@ class ComposableFilter(ABC):
         --------
         Dict
             A dictionary representing the result of the filtering operation.
-        """
+        """

tme/preprocessing/compose.py

@@ -39,12 +39,11 @@ class Compose:
 
         meta = self.transforms[0](**kwargs)
         for transform in self.transforms[1:]:
-
             kwargs.update(meta)
             ret = transform(**kwargs)
 
             if ret.get("is_multiplicative_filter", False):
-                backend.multiply(ret["data"], meta["data"], ret["data"])
+                backend.multiply(ret["data"], meta["data"], out=ret["data"])
                 ret["merge"] = None
 
             meta = ret