biotite 1.5.0__cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl

biotite/application/autodock/app.py

@@ -0,0 +1,500 @@
+# This source code is part of the Biotite package and is distributed
+# under the 3-Clause BSD License. Please see 'LICENSE.rst' for further
+# information.
+
+__name__ = "biotite.application.autodock"
+__author__ = "Patrick Kunzmann"
+__all__ = ["VinaApp"]
+
+import copy
+from tempfile import NamedTemporaryFile
+import numpy as np
+from biotite.application.application import AppState, requires_state
+from biotite.application.localapp import LocalApp, cleanup_tempfile
+from biotite.structure.bonds import find_connected
+from biotite.structure.error import BadStructureError
+from biotite.structure.io.pdbqt import PDBQTFile
+from biotite.structure.residues import get_residue_masks, get_residue_starts_for
+
+
+class VinaApp(LocalApp):
+    """
+    Dock a ligand to a receptor molecule using *AutoDock Vina*.
+
+    Parameters
+    ----------
+    ligand : AtomArray
+        The structure of the receptor molecule.
+        Must have an associated :class:`BondList`.
+        An associated ``charge`` annotation is recommended for proper
+        calculation of partial charges.
+    receptor : AtomArray, shape=(n,)
+        The structure of the receptor molecule.
+        Must have an associated :class:`BondList`.
+        An associated ``charge`` annotation is recommended for proper
+        calculation of partial charges.
+    center : ndarray, shape=(3,), dtype=float
+        The *xyz* coordinates for the center of the search space.
+    size : ndarray, shape=(3,), dtype=float
+        The size of the search space in *xyz* directions.
+    flexible : ndarray, shape=(n,), dtype=bool, optional
+        A boolean mask that indicates flexible amino acid side chains
+        in `receptor`.
+        Each residue, where at least one atom index is ``True`` in
+        `flexible`, is considered flexible.
+        By default, the receptor has no flexibility.
+    bin_path : str, optional
+        Path to the *Vina* binary.
+
+    Examples
+    --------
+
+    >>> # A dummy receptor and ligand
+    >>> ligand = residue("ASP")
+    >>> receptor = atom_array
+    >>> app = VinaApp(
+    ...     ligand, receptor,
+    ...     # Binding pocket is in the center of the receptor
+    ...     center=centroid(receptor),
+    ...     # 20 Å x 20 Å x 20 Å search space
+    ...     size=[20, 20, 20],
+    ...     # Handle residues 2 and 5 as flexible
+    ...     flexible=(receptor.res_id == 2) | (receptor.res_id == 5)
+    ... )
+    """
+
+    def __init__(self, ligand, receptor, center, size, flexible=None, bin_path="vina"):
+        super().__init__(bin_path)
+
+        if ligand.bonds is None:
+            raise ValueError("The ligand has no associated BondList")
+        if receptor.bonds is None:
+            raise ValueError("The receptor has no associated BondList")
+
+        self._ligand = ligand.copy()
+        self._receptor = receptor.copy()
+        self._center = copy.deepcopy(center)
+        self._size = copy.deepcopy(size)
+        self._is_flexible = flexible is not None
+        self._seed = None
+        self._exhaustiveness = None
+        self._number = None
+        self._energy_range = None
+
+        if self._is_flexible:
+            flexible_indices = np.where(flexible)[0]
+            self._flex_res_starts = np.unique(
+                get_residue_starts_for(receptor, flexible_indices)
+            )
+
+        self._ligand_file = NamedTemporaryFile("w", suffix=".pdbqt", delete=False)
+        self._receptor_file = NamedTemporaryFile("w", suffix=".pdbqt", delete=False)
+        self._receptor_flex_file = NamedTemporaryFile(
+            "w", suffix=".pdbqt", delete=False
+        )
+        self._out_file = NamedTemporaryFile("r", suffix=".pdbqt", delete=False)
+
+    @requires_state(AppState.CREATED)
+    def set_seed(self, seed):
+        """
+        Fix the seed for the random number generator to get
+        reproducible results.
+
+        By default, the seed is chosen randomly.
+
+        Parameters
+        ----------
+        seed : int
+            The seed for the random number generator.
+        """
+        self._seed = seed
+
+    @requires_state(AppState.CREATED)
+    def set_exhaustiveness(self, exhaustiveness):
+        """
+        Set the *exhaustiveness* parameter for *Vina*.
+
+        A higher exhaustiveness may lead to better docking results, but
+        also increases the computation time.
+        By default, the exhaustiveness is ``8``.
+
+        Parameters
+        ----------
+        exhaustiveness : int
+            The value for the exhaustiveness parameter.
+            Must be greater than 0.
+        """
+        self._exhaustiveness = exhaustiveness
+
+    @requires_state(AppState.CREATED)
+    def set_max_number_of_models(self, number):
+        """
+        Set the maximum number of binding modes to generate.
+
+        *Vina* may generate less modes, if the docking process does
+        not find enough distinct conformations.
+        By default, the maximum number is ``9``.
+
+        Parameters
+        ----------
+        number : int
+            The maximum number of generated modes/models.
+        """
+        self._number = number
+
+    @requires_state(AppState.CREATED)
+    def set_energy_range(self, energy_range):
+        """
+        Set the maximum energy range of the generated models.
+
+        *Vina* will ignore binding modes if the difference between this
+        mode and the best mode is greater than this value.
+        By default, the range is ``3.0``.
+
+        Parameters
+        ----------
+        energy_range : float
+            The energy range (kcal/mol).
+        """
+        self._energy_range = energy_range
+
+    def run(self):
+        # Use different atom ID ranges for atoms in ligand and receptor
+        # for unambiguous assignment, if the receptor contains flexible
+        # residues
+        self._ligand.set_annotation(
+            "atom_id", np.arange(1, self._ligand.array_length() + 1)
+        )
+        self._receptor.set_annotation(
+            "atom_id",
+            np.arange(
+                self._ligand.array_length() + 1,
+                self._ligand.array_length() + self._receptor.array_length() + 1,
+            ),
+        )
+
+        ligand_file = PDBQTFile()
+        # Contains 'true' entries for all atoms that have not been
+        # removed from ligand
+        self._ligand_mask = ligand_file.set_structure(
+            self._ligand, rotatable_bonds="all"
+        )
+        ligand_file.write(self._ligand_file)
+        self._ligand_file.flush()
+
+        if self._is_flexible:
+            self._rigid_mask = np.ones(self._receptor.array_length(), dtype=bool)
+            # Contains 'true' entries for all atoms that have not been
+            # removed from receptor in flexible side chains
+            self._receptor_mask = np.zeros(self._receptor.array_length(), dtype=bool)
+            for i, start in enumerate(self._flex_res_starts):
+                flex_mask, rigid_mask, root = self._get_flexible_residue(start)
+                self._rigid_mask &= rigid_mask
+                root_in_flex_residue = np.where(
+                    np.arange(self._receptor.array_length())[flex_mask] == root
+                )[0][0]
+                flex_file = PDBQTFile()
+                self._receptor_mask[flex_mask] |= flex_file.set_structure(
+                    self._receptor[flex_mask],
+                    rotatable_bonds="all",
+                    root=root_in_flex_residue,
+                    include_torsdof=False,
+                )
+                # Enclose each flexible residue
+                # with BEGIN_RES and END_RES
+                self._receptor_flex_file.write(f"BEGIN_RES {i}\n")
+                flex_file.write(self._receptor_flex_file)
+                self._receptor_flex_file.write(f"END_RES {i}\n")
+            self._receptor_flex_file.flush()
+
+            receptor_file = PDBQTFile()
+            receptor_file.set_structure(
+                self._receptor[self._rigid_mask],
+                rotatable_bonds=None,
+                include_torsdof=False,
+            )
+            receptor_file.write(self._receptor_file)
+            self._receptor_file.flush()
+
+        else:
+            receptor_file = PDBQTFile()
+            receptor_file.set_structure(
+                self._receptor, rotatable_bonds=None, include_torsdof=False
+            )
+            receptor_file.write(self._receptor_file)
+            self._receptor_file.flush()
+
+        arguments = [
+            "--ligand",
+            self._ligand_file.name,
+            "--receptor",
+            self._receptor_file.name,
+            "--out",
+            self._out_file.name,
+            "--center_x",
+            f"{self._center[0]:.3f}",
+            "--center_y",
+            f"{self._center[1]:.3f}",
+            "--center_z",
+            f"{self._center[2]:.3f}",
+            "--size_x",
+            f"{self._size[0]:.3f}",
+            "--size_y",
+            f"{self._size[1]:.3f}",
+            "--size_z",
+            f"{self._size[2]:.3f}",
+        ]
+        if self._seed is not None:
+            arguments.extend(["--seed", str(self._seed)])
+        if self._exhaustiveness is not None:
+            arguments.extend(["--exhaustiveness", str(self._exhaustiveness)])
+        if self._number is not None:
+            arguments.extend(["--num_modes", str(self._number)])
+        if self._energy_range is not None:
+            arguments.extend(["--energy_range", str(self._energy_range)])
+        if self._is_flexible:
+            arguments.extend(["--flex", str(self._receptor_flex_file.name)])
+
+        self.set_arguments(arguments)
+        super().run()
+
+    def evaluate(self):
+        super().evaluate()
+        out_file = PDBQTFile.read(self._out_file)
+
+        models = out_file.get_structure()
+
+        n_ligand_atoms = np.count_nonzero(self._ligand_mask)
+        self._ligand_models = models[..., :n_ligand_atoms]
+        self._flex_models = models[..., n_ligand_atoms:]
+        self._n_models = models.stack_depth()
+
+        remarks = out_file.get_remarks()
+        self._energies = np.array(
+            # VINA RESULT:      -5.8      0.000      0.000
+            #                     ^
+            [float(remark[12:].split()[0]) for remark in remarks]
+        )
+
+    def clean_up(self):
+        super().clean_up()
+        cleanup_tempfile(self._ligand_file)
+        cleanup_tempfile(self._receptor_file)
+        cleanup_tempfile(self._receptor_flex_file)
+        cleanup_tempfile(self._out_file)
+
+    @requires_state(AppState.JOINED)
+    def get_energies(self):
+        """
+        Get the predicted binding energy for each generated binding
+        mode.
+
+        Returns
+        -------
+        energies : ndarray, dtype=float
+            The predicted binding energies (kcal/mol).
+            The energies are sorted from best to worst.
+        """
+        return self._energies
+
+    @requires_state(AppState.JOINED)
+    def get_ligand_models(self):
+        """
+        Get the ligand structure with the conformations for each
+        generated binding mode.
+
+        Returns
+        -------
+        ligand : AtomArrayStack
+            The docked ligand.
+            Each model corresponds to one binding mode.
+            The models are sorted from best to worst predicted binding
+            affinity.
+
+        Notes
+        -----
+        The returned structure may contain less atoms than the input
+        structure, as *Vina* removes nonpolar hydrogen atoms.
+        Furthermore, the returned structure contains *AutoDock* atom
+        types as ``element`` annotation.
+        """
+        return self._ligand_models
+
+    @requires_state(AppState.JOINED)
+    def get_ligand_coord(self):
+        """
+        Get the ligand coordinates for each generated binding mode.
+
+        Returns
+        -------
+        coord : ndarray, shape=(m,n,3), dtype=float
+            The coordinates for *m* binding modes and *n* atoms
+            of the input ligand.
+            The models are sorted from best to worst predicted binding
+            affinity.
+            Missing coordinates due to the removed nonpolar hydrogen
+            atoms are set to *NaN*.
+        """
+        coord = np.full(
+            (self._n_models, self._ligand.array_length(), 3), np.nan, dtype=np.float32
+        )
+        coord[:, self._ligand_mask] = self._ligand_models.coord
+        return coord
+
+    @requires_state(AppState.JOINED)
+    def get_flexible_residue_models(self):
+        """
+        Get the structure for the flexible side chains with the
+        conformations for each generated binding mode.
+
+        If no flexible side chains were defined, the returned
+        :class:`AtomArrayStack` contains no atoms.
+
+        Returns
+        -------
+        side_chains : AtomArrayStack
+            The docked side chains.
+            Each model corresponds to one binding mode.
+            The models are sorted from best to worst predicted binding
+            affinity.
+
+        Notes
+        -----
+        The returned structure may contain less atoms than the input
+        structure, as *Vina* removes nonpolar hydrogen atoms.
+        Furthermore, the returned structure contains *AutoDock* atom
+        types as ``element`` annotation.
+        """
+        return self._flex_models
+
+    @requires_state(AppState.JOINED)
+    def get_receptor_coord(self):
+        """
+        Get the get_receptor_coord coordinates for each generated
+        binding mode.
+
+        Returns
+        -------
+        coord : ndarray, shape=(m,n,3), dtype=float
+            The coordinates for *m* binding modes and *n* atoms
+            of the input receptor.
+            The models are sorted from best to worst predicted binding
+            affinity.
+            Missing coordinates due to the removed nonpolar hydrogen
+            atoms from flexible side chains are set to *NaN*.
+
+        Notes
+        -----
+        The output is only meaningful, if flexible side chains were
+        defined.
+        Otherwise, the returned coordinates are simply *m* repetitions
+        of the input receptor coordinates.
+        """
+        coord = np.repeat(
+            self._receptor.coord[np.newaxis, ...], repeats=self._n_models, axis=0
+        )
+        if self._is_flexible:
+            # Replace original coordinates with modeled coordinates
+            # for the the flexible side chains
+            # The coordinates from removed atoms are NaN
+            coord[:, ~self._rigid_mask] = np.nan
+            coord[:, self._receptor_mask] = self._flex_models.coord
+        return coord
+
+    def _get_flexible_residue(self, residue_start):
+        residue_indices = np.where(
+            get_residue_masks(self._receptor, [residue_start])[0]
+        )[0]
+        root_indices_in_residue = np.isin(
+            self._receptor.atom_name[residue_indices], ("CA",)
+        )
+        root_indices = residue_indices[root_indices_in_residue]
+        if len(root_indices) == 0:
+            raise BadStructureError("Found no CA atom in residue")
+        if len(root_indices) > 1:
+            raise BadStructureError("Multiple CA atom in residue")
+        root_index = root_indices[0]
+
+        # Find the index of the atom connected to root on the flexible
+        # side chain (CB)
+        root_connect_indices, _ = self._receptor.bonds.get_bonds(root_index)
+        connected_index = None
+        try:
+            connected_index = root_connect_indices[
+                np.isin(self._receptor.atom_name[root_connect_indices], ("CB",))
+            ][0]
+        except IndexError:
+            # Residue has no appropriate connection (e.g. in glycine)
+            # -> There is no atom in the flexible side chain
+            flex_mask = np.zeros(self._receptor.array_length(), dtype=bool)
+            rigid_mask = np.ones(self._receptor.array_length(), dtype=bool)
+            return flex_mask, rigid_mask, root_index
+
+        # Remove the root bond from the bond list
+        # to find the atoms involved in the flexible part
+        bonds = self._receptor.bonds.copy()
+        bonds.remove_bond(root_index, connected_index)
+        flexible_indices = find_connected(bonds, connected_index)
+        if root_index in flexible_indices:
+            raise BadStructureError(
+                "There are multiple connections between the flexible and "
+                "rigid part, maybe a cyclic residue like proline was selected"
+            )
+
+        flex_mask = np.zeros(self._receptor.array_length(), dtype=bool)
+        flex_mask[flexible_indices] = True
+        rigid_mask = ~flex_mask
+        # Root index is part of rigid and flexible part
+        flex_mask[root_index] = True
+
+        return flex_mask, rigid_mask, root_index
+
+    @staticmethod
+    def dock(ligand, receptor, center, size, flexible=None, bin_path="vina"):
+        """
+        Dock a ligand to a receptor molecule using *AutoDock Vina*.
+
+        This is a convenience function, that wraps the :class:`VinaApp`
+        execution.
+
+        Parameters
+        ----------
+        ligand : AtomArray
+            The structure of the receptor molecule.
+            Must have an associated :class:`BondList`.
+            An associated ``charge`` annotation is recommended for proper
+            calculation of partial charges.
+        receptor : AtomArray, shape=(n,)
+            The structure of the receptor molecule.
+            Must have an associated :class:`BondList`.
+            An associated ``charge`` annotation is recommended for proper
+            calculation of partial charges.
+        center : ndarray, shape=(3,), dtype=float
+            The *xyz* coordinates for the center of the search space.
+        size : ndarray, shape=(3,), dtype=float
+            The size of the search space in *xyz* directions.
+        flexible : ndarray, shape=(n,), dtype=bool, optional
+            A boolean mask that indicates flexible amino acid side chains
+            in `receptor`.
+            Each residue, where at least one atom index is ``True`` in
+            `flexible`, is considered flexible.
+            By default, the receptor has no flexibility.
+        bin_path : str, optional
+            Path to the *Vina* binary.
+
+        Returns
+        -------
+        coord : ndarray, shape=(m,n,3), dtype=float
+            The docked ligand coordinates for *m* binding modes and
+            *n* atoms of the input ligand.
+            The models are sorted from best to worst predicted binding
+            affinity.
+            Missing coordinates due to the removed nonpolar hydrogen
+            atoms are set to *NaN*.
+        energies : ndarray, shape=(m,), dtype=float
+            The corresponding predicted binding energies (kcal/mol).
+        """
+        app = VinaApp(ligand, receptor, center, size, flexible, bin_path)
+        app.start()
+        app.join()
+        return app.get_ligand_coord(), app.get_energies()

biotite/application/blast/__init__.py

@@ -0,0 +1,14 @@
+# This source code is part of the Biotite package and is distributed
+# under the 3-Clause BSD License. Please see 'LICENSE.rst' for further
+# information.
+
+"""
+A subpackage for heuristic local alignments against a large database
+using BLAST.
+"""
+
+__name__ = "biotite.application.blast"
+__author__ = "Patrick Kunzmann"
+
+from .alignment import *
+from .webapp import *

biotite/application/blast/alignment.py

@@ -0,0 +1,92 @@
+# This source code is part of the Biotite package and is distributed
+# under the 3-Clause BSD License. Please see 'LICENSE.rst' for further
+# information.
+
+__name__ = "biotite.application.blast"
+__author__ = "Patrick Kunzmann"
+__all__ = ["BlastAlignment"]
+
+from biotite.sequence.align.alignment import Alignment
+
+
+class BlastAlignment(Alignment):
+    """
+    A specialized :class:`Alignment` class for alignments using the
+    BLAST application. It stores additional data, like the E-value,
+    the HSP position and a description of the hit sequence.
+
+    Like its superclass, all attributes of a :class:`BlastAlignment` are
+    public. The attributes are the same as the constructor parameters.
+
+    Parameters
+    ----------
+    sequences : list
+        A list of aligned sequences. Does actually not contain the
+        complete original sequences, but the HSP sequences.
+    trace : ndarray, dtype=int, shape=(n,m)
+        The alignment trace.
+    score : int
+        Alignment score.
+    e_value : float
+        Expectation value for the number of random sequences of a
+        similar sized database getting an equal or higher score by
+        change when aligned with the query sequence.
+    query_interval : tuple of int
+        Describes the position of the HSP part of the query sequence
+        in the original query sequence. The first element is the start
+        position, the second element is the inclusive stop position.
+        Indexing starts at 1.
+    hit_interval : tuple of int
+        Analogous to `query_interval`, this describes the position of
+        the HSP part of the hit sequence in the complete hit sequence.
+    hit_id : str
+        The NCBI *unique identifier* (UID) of the hit sequence.
+    hit_definition : str
+        The name of the hit sequence.
+    """
+
+    def __init__(
+        self,
+        sequences,
+        trace,
+        score,
+        e_value,
+        query_interval,
+        hit_interval,
+        hit_id,
+        hit_definition,
+    ):
+        super().__init__(sequences, trace, score)
+        self.e_value = e_value
+        self.query_interval = query_interval
+        self.hit_interval = hit_interval
+        self.hit_id = hit_id
+        self.hit_definition = hit_definition
+
+    def __eq__(self, item):
+        if not isinstance(item, BlastAlignment):
+            return False
+        if self.e_value != item.e_value:
+            return False
+        if self.query_interval != item.query_interval:
+            return False
+        if self.hit_interval != item.hit_interval:
+            return False
+        if self.hit_id != item.hit_id:
+            return False
+        if self.hit_definition != item.hit_definition:
+            return False
+        return super().__eq__(item)
+
+    def __getitem__(self, index):
+        super_alignment = super().__getitem__(index)
+        return BlastAlignment(
+            super_alignment.sequences,
+            super_alignment.trace,
+            super_alignment.score,
+            self.e_value,
+            self.query_interval,
+            self.hit_interval,
+            self.hit_id,
+            self.hit_definition,
+        )