biotite 1.5.0__cp314-cp314-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl

biotite/database/rcsb/query.py

@@ -0,0 +1,963 @@
+# 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.database.rcsb"
+__author__ = "Patrick Kunzmann, Maximilian Dombrowsky"
+__all__ = [
+    "Query",
+    "SingleQuery",
+    "CompositeQuery",
+    "BasicQuery",
+    "FieldQuery",
+    "SequenceQuery",
+    "StructureQuery",
+    "MotifQuery",
+    "Sorting",
+    "Grouping",
+    "DepositGrouping",
+    "IdentityGrouping",
+    "UniprotGrouping",
+    "search",
+    "count",
+]
+
+import abc
+import copy
+import json
+from datetime import datetime
+import numpy as np
+import requests
+from biotite.database.error import RequestError
+from biotite.sequence.seqtypes import NucleotideSequence
+
+_search_url = "https://search.rcsb.org/rcsbsearch/v2/query"
+_scope_to_target = {
+    "protein": "pdb_protein_sequence",
+    "rna": "pdb_rna_sequence",
+    "dna": "pdb_dna_sequence",
+}
+
+
+class Query(metaclass=abc.ABCMeta):
+    """
+    A representation of a JSON query for the RCSB search API.
+
+    This is the abstract base class for all queries.
+    """
+
+    @abc.abstractmethod
+    def get_content(self):
+        """
+        Get the query content, i.e. the data belonging to the
+        ``'query'`` attribute in the RCSB search API.
+
+        This content is converted into JSON by the :func:`search`
+        and :func:`count` functions.
+
+        Returns
+        -------
+        content : dict
+            The content dictionary for the ``'query'`` attributes.
+        """
+        pass
+
+    def __and__(self, query):
+        return CompositeQuery([self, query], "and")
+
+    def __or__(self, query):
+        return CompositeQuery([self, query], "or")
+
+
+class SingleQuery(Query, metaclass=abc.ABCMeta):
+    """
+    A terminal query node for the RCSB search API.
+
+    Multiple :class:`SingleQuery` objects can be combined to
+    :class:`CompositeQuery` objects using the ``|`` and ``&`` operators.
+
+    This is the abstract base class for all queries that are
+    terminal nodes.
+    """
+
+    @abc.abstractmethod
+    def get_content(self):
+        return {"parameters": {}}
+
+
+class CompositeQuery(Query):
+    """
+    A group query node for the RCSB search API.
+
+    A composite query is an combination of other queries, combined
+    either with the `'and'` or `'or'` operator.
+    Usually, a :class:`CompositeQuery` will not be created by calling
+    its constructor, but by combining queries using the ``|`` or ``&``
+    operator.
+
+    Parameters
+    ----------
+    queries : iterable object of Query
+        The queries to be combined.
+    operator : {'or', 'and'}
+        The type of combination.
+    """
+
+    def __init__(self, queries, operator):
+        self._queries = queries
+        if operator not in ("or", "and"):
+            raise ValueError(f"Operator must be 'or' or 'and', not '{operator}'")
+        self._operator = operator
+
+    def get_content(self):
+        """
+        A dictionary representation of the query.
+        This dictionary is the content of the ``'query'`` key in the
+        JSON query.
+
+        Returns
+        -------
+        content : dict
+            The dictionary representation of the query.
+        """
+        content = {
+            "type": "group",
+            "logical_operator": self._operator,
+            "nodes": [query.get_content() for query in self._queries],
+        }
+        return content
+
+
+class BasicQuery(SingleQuery):
+    """
+    A text query for searching for a given term across all available
+    fields.
+
+    Parameters
+    ----------
+    term : str
+        The search term.
+        If the term contains multiple words, the query will return
+        results where the entire term is present.
+        The matching is not case-sensitive.
+        Logic combinations of terms is described
+        `here <https://search.rcsb.org/#basic-queries>`_.
+
+    Examples
+    --------
+
+    >>> query = BasicQuery("Miniprotein Construct")
+    >>> print(sorted(search(query)))
+    ['1L2Y']
+    """
+
+    def __init__(self, term):
+        super().__init__()
+        self._term = term
+
+    def get_content(self):
+        content = super().get_content()
+        content["type"] = "terminal"
+        content["service"] = "full_text"
+        content["parameters"]["value"] = f'"{self._term}"'
+        return content
+
+
+class FieldQuery(SingleQuery):
+    """
+    A text query for searching for values in a given field using the
+    given operator.
+
+    The operators are keyword arguments of this function and the search
+    value is the value given to the respective parameter.
+    The operators are mutually exclusive.
+    If none is given, the search will return results where the given
+    field exists.
+
+    A :class:`FieldQuery` is negated using the ``~`` operator.
+
+    Parameters
+    ----------
+    field : str
+        The field to search in.
+    molecular_definition : bool, optional
+        If set true, this query searches in fields
+        associated with
+        `molecular definitions <https://search.rcsb.org/chemical-search-attributes.html>`_.
+        If false (default), this query searches in fields
+        associated with `PDB structures <https://search.rcsb.org/structure-search-attributes.html>`_.
+    case_sensitive : bool, optional
+        If set to true, searches are case sensitive.
+        By default matching is case-insensitive.
+    exact_match : str, optional
+        Operator for returning results whose field exactly matches the
+        given value.
+    contains_words, contains_phrase : str, optional
+        Operator for returning results whose field matches
+        individual words from the given value or the value as exact
+        phrase, respectively.
+    greater, less, greater_or_equal, less_or_equal, equals : int or float or datetime, optional
+        Operator for returning results whose field values are larger,
+        smaller or equal to the given value.
+    range, range_closed : tuple(int, int) or tuple(float, float) or tuple(datetime, datetime), optional
+        Operator for returning results whose field matches values within
+        the given range.
+        `range_closed` includes the interval limits.
+    is_in : tuple of str or list of str, optional
+        Operator for returning results whose field matches any of the
+        values in the given list.
+
+    Notes
+    -----
+    A complete list of the available fields and its supported operators
+    is documented at
+    `<https://search.rcsb.org/structure-search-attributes.html>`_
+    and
+    `<https://search.rcsb.org/chemical-search-attributes.html>`_.
+
+    Examples
+    --------
+
+    >>> query = FieldQuery("reflns.d_resolution_high", less_or_equal=0.6)
+    >>> print(sorted(search(query)))
+    ['1EJG', '1I0T', '3NIR', '3P4J', '4JLJ', '5D8V', '5NW3', '7ATG', '7R0H']
+    """
+
+    def __init__(
+        self, field, molecular_definition=False, case_sensitive=False, **kwargs
+    ):
+        super().__init__()
+        self._negation = False
+        self._field = field
+        self._mol_definition = molecular_definition
+        self._case_sensitive = case_sensitive
+
+        if len(kwargs) > 1:
+            raise TypeError("Only one operator must be given")
+        elif len(kwargs) == 1:
+            self._operator = list(kwargs.keys())[0]
+            self._value = list(kwargs.values())[0]
+        else:
+            # No operator is given
+            self._operator = "exists"
+            self._value = None
+
+        if self._operator not in [
+            "exact_match",
+            "contains_words",
+            "contains_phrase",
+            "greater",
+            "less",
+            "greater_or_equal",
+            "less_or_equal",
+            "equals",
+            "range",
+            "range_closed",
+            "is_in",
+            "exists",
+        ]:
+            raise TypeError(
+                f"Constructor got an unexpected keyword argument '{self._operator}'"
+            )
+
+        # Convert dates into ISO 8601
+        if isinstance(self._value, datetime):
+            self._value = _to_isoformat(self._value)
+        elif isinstance(self._value, (tuple, list, np.ndarray)):
+            self._value = [
+                _to_isoformat(val) if isinstance(val, datetime) else val
+                for val in self._value
+            ]
+
+        # Create dictionary for 'range' operator
+        if self._operator == "range":
+            self._value = {
+                "from": self._value[0],
+                "include_lower": False,
+                "to": self._value[1],
+                "include_upper": False,
+            }
+        elif self._operator == "range_closed":
+            self._value = {
+                "from": self._value[0],
+                "include_lower": True,
+                "to": self._value[1],
+                "include_upper": True,
+            }
+
+        # Rename operators to names used in API
+        if self._operator == "is_in":
+            # 'in' is not an available parameter name in Python
+            self._operator = "in"
+        elif self._operator == "range_closed":
+            # For backwards compatibility
+            self._operator = "range"
+
+    def get_content(self):
+        content = super().get_content()
+        content["type"] = "terminal"
+        if self._mol_definition:
+            content["service"] = "text_chem"
+        else:
+            content["service"] = "text"
+        content["parameters"]["attribute"] = self._field
+        content["parameters"]["operator"] = self._operator
+        content["parameters"]["negation"] = self._negation
+        content["parameters"]["case_sensitive"] = self._case_sensitive
+        if self._value is not None:
+            content["parameters"]["value"] = self._value
+        return content
+
+    def __invert__(self):
+        clone = copy.deepcopy(self)
+        clone._negation = not clone._negation
+        return clone
+
+
+class SequenceQuery(SingleQuery):
+    """
+    A query for protein/DNA/RNA molecules with a sequence similar to a
+    given input sequence using
+    `MMseqs2 <https://github.com/soedinglab/mmseqs2>`_.
+
+    Parameters
+    ----------
+    sequence : Sequence or str
+        The input sequence.
+        If `sequence` is a :class:`NucleotideSequence` and the `scope`
+        is ``'rna'``, ``'T'`` is automatically replaced by ``'U'``.
+    scope : {'protein', 'dna', 'rna'}
+        The type of molecule to find.
+    min_identity : float, optional
+        A match is only returned, if the sequence identity between
+        the match and the input sequence exceeds this value.
+        Must be between 0 and 1.
+        By default, the sequence identity is ignored.
+    max_expect_value : float, optional
+        A match is only returned, if the *expect value* (E-value) does
+        not exceed this value.
+        By default, the value is effectively ignored.
+
+    Notes
+    -----
+    *MMseqs2* is run on the RCSB servers.
+
+    Examples
+    --------
+
+    >>> sequence = "NLYIQWLKDGGPSSGRPPPS"
+    >>> query = SequenceQuery(sequence, scope="protein", min_identity=0.95)
+    >>> print(sorted(search(query)))
+    ['1L2Y', '2LDJ', '9G22', '9G2N', '9G2O', '9G31', '9G32', '9GDL', '9GDN', '9GDT', '9GDU', '9GE1']
+    """
+
+    def __init__(self, sequence, scope, min_identity=0.0, max_expect_value=10000000.0):
+        super().__init__()
+        self._target = _scope_to_target.get(scope.lower())
+        if self._target is None:
+            raise ValueError(f"'{scope}' is an invalid scope")
+
+        if isinstance(sequence, NucleotideSequence) and scope.lower() == "rna":
+            self._sequence = str(sequence).replace("T", "U")
+        else:
+            self._sequence = str(sequence)
+
+        self._min_identity = min_identity
+        self._max_expect_value = max_expect_value
+
+    def get_content(self):
+        content = super().get_content()
+        content["type"] = "terminal"
+        content["service"] = "sequence"
+        content["parameters"]["value"] = self._sequence
+        content["parameters"]["target"] = self._target
+        content["parameters"]["identity_cutoff"] = self._min_identity
+        content["parameters"]["evalue_cutoff"] = self._max_expect_value
+        return content
+
+
+class MotifQuery(SingleQuery):
+    """
+    A query for protein/DNA/RNA molecules containing the given sequence
+    motif.
+
+    Parameters
+    ----------
+    pattern : str
+        The sequence pattern.
+    pattern_type : {'simple', 'prosite', 'regex'}
+        The type of the pattern.
+    scope : {'protein', 'dna', 'rna'}
+        The type of molecule to find.
+
+    Examples
+    --------
+
+    >>> query = MotifQuery(
+    ...     "C-x(2,4)-C-x(3)-[LIVMFYWC]-x(8)-H-x(3,5)-H.",
+    ...     "prosite",
+    ...     "protein"
+    ... )
+    """
+
+    def __init__(self, pattern, pattern_type, scope):
+        super().__init__()
+        self._pattern = pattern
+        self._pattern_type = pattern_type
+        self._target = _scope_to_target.get(scope.lower())
+
+    def get_content(self):
+        content = super().get_content()
+        content["type"] = "terminal"
+        content["service"] = "seqmotif"
+        content["parameters"]["value"] = self._pattern
+        content["parameters"]["pattern_type"] = self._pattern_type
+        content["parameters"]["target"] = self._target
+        return content
+
+
+class StructureQuery(SingleQuery):
+    """
+    A query for protein/DNA/RNA molecules with structural similarity
+    to the query structure.
+
+    Either the chain or assembly ID of the query structure must be
+    specified.
+
+    Parameters
+    ----------
+    pdb_id : str
+        The PDB ID of the query structure.
+    chain : str, optional
+        The chain ID (more exactly ``asym_id``) of the query structure.
+    assembly : str, optional
+        The assembly ID (``assembly_id``) of the query structure.
+    strict : bool, optional
+        If true, structure comparison is strict, otherwise it is
+        relaxed.
+
+    Examples
+    --------
+
+    >>> query = StructureQuery("1L2Y", chain="A")
+    >>> print(sorted(search(query)))
+    ['1L2Y', '1RIJ', '2JOF', '2LDJ', '2M7D', '7MQS', '9DPF']
+    """
+
+    def __init__(self, pdb_id, chain=None, assembly=None, strict=True):
+        super().__init__()
+
+        if (chain is None and assembly is None) or (
+            chain is not None and assembly is not None
+        ):
+            raise TypeError("Either the chain ID or assembly ID must be set")
+        elif chain is None:
+            self._value = {"entry_id": pdb_id, "asssembly_id": assembly}
+        else:
+            self._value = {"entry_id": pdb_id, "asym_id": chain}
+
+        self._operator = "strict_shape_match" if strict else "relaxed_shape_match"
+
+    def get_content(self):
+        content = super().get_content()
+        content["type"] = "terminal"
+        content["service"] = "structure"
+        content["parameters"]["value"] = self._value
+        content["parameters"]["operator"] = self._operator
+        return content
+
+
+class Sorting:
+    def __init__(self, field, descending=True):
+        self._field = field
+        self._descending = descending
+
+    @property
+    def field(self):
+        return self._field
+
+    @property
+    def descending(self):
+        return self._descending
+
+    def get_content(self):
+        """
+        Get the sorting content, i.e. the data belonging to the
+        ``'sort'`` and ``'ranking_criteria_type'`` attributes in the
+        RCSB search API.
+
+        This content is converted into JSON by the :func:`search`
+        function.
+
+        Returns
+        -------
+        content : dict
+            The content dictionary for the ``'sort'`` and
+            ``'ranking_criteria_type'`` attributes.
+        """
+        direction = "desc" if self._descending else "asc"
+        return {"sort_by": self._field, "direction": direction}
+
+
+class Grouping(metaclass=abc.ABCMeta):
+    """
+    A representation of the JSON grouping options of the RCSB search
+    API.
+
+    Parameters
+    ----------
+    sort_by : str or Sorting, optional
+        If specified, the returned PDB IDs within each group are sorted
+        by the values of the given field name.
+        A complete list of the available fields is documented at
+        `<https://search.rcsb.org/structure-search-attributes.html>`_.
+        and
+        `<https://search.rcsb.org/chemical-search-attributes.html>`_.
+        If a string is given, sorting is performed in descending order.
+        To choose the order a :class:`Sorting` object needs to be
+        provided.
+    """
+
+    def __init__(self, sort_by=None):
+        if sort_by is None:
+            self._sorting = None
+        elif isinstance(sort_by, Sorting):
+            self._sorting = sort_by
+        else:
+            self._sorting = Sorting(sort_by)
+
+    @abc.abstractmethod
+    def get_content(self):
+        """
+        Get the grouping content, i.e. the data belonging to the
+        ``'group_by'`` attribute in the RCSB search API.
+
+        This content is converted into JSON by the :func:`search`
+        and :func:`count` functions.
+
+        ABSTRACT: Override when inheriting.
+
+        Returns
+        -------
+        content : dict
+            The content dictionary for the ``'group_by'`` attributes.
+        """
+        if self._sorting is not None:
+            return {"ranking_criteria_type": self._sorting.get_content()}
+        else:
+            return {}
+
+    @abc.abstractmethod
+    def is_compatible_return_type(self, return_type):
+        """
+        Check whether this :class:`Group` is compatible with the
+        RCSB search API ``return_type``.
+
+        ABSTRACT: Override when inheriting.
+
+        Parameters
+        ----------
+        return_type : str
+            The ``return_type`` attribute to be checked.
+
+        Returns
+        -------
+        is_compatible : bool
+            True, if this :class:`Group` is compatible with the
+            `return_type`, false otherwise.
+        """
+        pass
+
+
+class DepositGrouping(Grouping):
+    """
+    This class groups PDB entries if they were deposited as a
+    collection.
+    Such a group usually contain the same protein with e.g. a different
+    bound molecule.
+
+    This :class:`Grouping` is only applicable, if the
+    :func:`count()`/:func:`search()` return type is set to ``entry``.
+
+    Parameters
+    ----------
+    sort_by : str or Sorting, optional
+        If specified, the returned PDB IDs within each group are sorted
+        by the values of the given field name.
+        A complete list of the available fields is documented at
+        `<https://search.rcsb.org/structure-search-attributes.html>`_.
+        and
+        `<https://search.rcsb.org/chemical-search-attributes.html>`_.
+        If a string is given, sorting is performed in descending order.
+        To choose the order a :class:`Sorting` object needs to be
+        provided.
+    """
+
+    def get_content(self):
+        content = super().get_content()
+        content["aggregation_method"] = "matching_deposit_group_id"
+        return content
+
+    def is_compatible_return_type(self, return_type):
+        return return_type == "entry"
+
+
+class IdentityGrouping(Grouping):
+    """
+    This class groups protein chains with a given sequence identity
+    with each other.
+
+    This :class:`Grouping` is only applicable, if the
+    :func:`count()`/:func:`search()` return type is set to
+    ``polymer_entity``.
+
+    Parameters
+    ----------
+    similarity_cutoff : {100, 95, 90, 70, 50, 30}
+        The sequence identity in percent at which the structures are
+        grouped.
+        In other words, a returned group contains sequences that have
+        `similarity_cutoff` sequence identity with each other.
+        Since the PDB uses precalculated clusters, only certain values
+        are available.
+    sort_by : str or Sorting, optional
+        If specified, the returned PDB IDs within each group are sorted
+        by the values of the given field name.
+        A complete list of the available fields is documented at
+        `<https://search.rcsb.org/structure-search-attributes.html>`_.
+        and
+        `<https://search.rcsb.org/chemical-search-attributes.html>`_.
+        If a string is given, sorting is performed in descending order.
+        To choose the order a :class:`Sorting` object needs to be
+        provided.
+    """
+
+    def __init__(self, similarity_cutoff, sort_by=None):
+        super().__init__(sort_by)
+        if similarity_cutoff not in (100, 95, 90, 70, 50, 30):
+            raise ValueError(
+                f"A similarity cutoff of {similarity_cutoff}% is not supported"
+            )
+        self._similarity_cutoff = similarity_cutoff
+
+    def get_content(self):
+        content = super().get_content()
+        content["aggregation_method"] = "sequence_identity"
+        content["similarity_cutoff"] = self._similarity_cutoff
+        return content
+
+    def is_compatible_return_type(self, return_type):
+        return return_type == "polymer_entity"
+
+
+class UniprotGrouping(Grouping):
+    """
+    This class groups protein chains that point to the same *Uniprot*
+    accession ID.
+
+    This :class:`Grouping` is only applicable, if the
+    :func:`count()`/:func:`search()` return type is set to
+    ``polymer_entity``.
+
+    Parameters
+    ----------
+    sort_by : str or Sorting, optional
+        If specified, the returned PDB IDs within each group are sorted
+        by the values of the given field name.
+        A complete list of the available fields is documented at
+        `<https://search.rcsb.org/structure-search-attributes.html>`_.
+        and
+        `<https://search.rcsb.org/chemical-search-attributes.html>`_.
+        If a string is given, sorting is performed in descending order.
+        To choose the order a :class:`Sorting` object needs to be
+        provided.
+    """
+
+    def get_content(self):
+        content = super().get_content()
+        content["aggregation_method"] = "matching_uniprot_accession"
+        return content
+
+    def is_compatible_return_type(self, return_type):
+        return return_type == "polymer_entity"
+
+
+def count(query, return_type="entry", group_by=None, content_types=("experimental",)):
+    """
+    Count PDB entries that meet the given query requirements,
+    via the RCSB search API.
+
+    This function requires an internet connection.
+
+    Parameters
+    ----------
+    query : Query
+        The search query.
+    return_type : {'entry', 'assembly', 'polymer_entity', 'non_polymer_entity', 'polymer_instance'}, optional
+        The type of the counted identifiers:
+
+        - ``'entry'``: All macthing PDB entries are counted.
+        - ``'assembly'``: All matching assemblies are counted.
+        - ``'polymer_entity'``: All matching polymeric entities are
+          counted.
+        - ``'non_polymer_entity'``: All matching non-polymeric entities
+          are counted.
+        - ``'polymer_instance'``: All matching chains are counted.
+    group_by : Grouping
+        If this parameter is set, the number of groups is returned
+        instead.
+    content_types : iterable of {"experimental", "computational"}, optional
+        Specify whether experimental and computational structures should
+        be included.
+        At least one of them needs to be specified.
+        By default only experimental structures are included.
+        Note, that identifiers for computational structures cannot be
+        downloaded via :func:`biotite.database.rcsb.fetch()` as they
+        point to *AlphaFold DB* and *ModelArchive*.
+
+    Returns
+    -------
+    count : int
+        The total number of PDB IDs (or groups) that would be returned
+        by calling :func:`search()` using the same parameters.
+
+    Notes
+    -----
+    If `group_by` is set, the number of results may be lower than in an
+    ungrouped query, as grouping is not applicable to all structures.
+    For example a DNA structure has no associated *Uniprot* accession
+    and hence is omitted by :class:`UniprotGrouping`.
+
+    Examples
+    --------
+
+    >>> query = FieldQuery("reflns.d_resolution_high", less_or_equal=0.6)
+    >>> print(count(query))
+    9
+    >>> ids = search(query)
+    >>> print(sorted(ids))
+    ['1EJG', '1I0T', '3NIR', '3P4J', '4JLJ', '5D8V', '5NW3', '7ATG', '7R0H']
+    """
+    query_dict = _initialize_query_dict(query, return_type, group_by, content_types)
+
+    query_dict["request_options"]["return_counts"] = True
+
+    r = requests.get(_search_url, params={"json": json.dumps(query_dict)})
+
+    if r.status_code == 200:
+        if group_by is None:
+            return r.json()["total_count"]
+        else:
+            return r.json()["group_by_count"]
+    elif r.status_code == 204:
+        # Search did not return any results
+        return 0
+    else:
+        try:
+            raise RequestError(f"Error {r.status_code}: {r.json()['message']}")
+        except json.decoder.JSONDecodeError:
+            # In case there an error response without message
+            raise RequestError(f"Error {r.status_code}")
+
+
+def search(
+    query,
+    return_type="entry",
+    range=None,
+    sort_by=None,
+    group_by=None,
+    return_groups=False,
+    content_types=("experimental",),
+):
+    """
+    Get all PDB IDs that meet the given query requirements,
+    via the RCSB search API.
+
+    This function requires an internet connection.
+
+    Parameters
+    ----------
+    query : Query
+        The search query.
+    return_type : {'entry', 'assembly', 'polymer_entity', 'non_polymer_entity', 'polymer_instance'}, optional
+        The type of the returned identifiers:
+
+        - ``'entry'``: Only the PDB ID is returned (e.g. ``'XXXX'``).
+          These can be used directly as input to :func:`fetch()`.
+        - ``'assembly'``: The PDB ID appended with assembly ID is
+          returned (e.g. ``'XXXX-1'``).
+        - ``'polymer_entity'``: The PDB ID appended with entity ID of
+          polymers is returned (e.g. ``'XXXX_1'``).
+        - ``'non_polymer_entity'``: The PDB ID appended with entity ID
+          of non-polymeric entities is returned (e.g. ``'XXXX_1'``).
+        - ``'polymer_instance'``: The PDB ID appended with chain ID
+          (more exactly ``'asym_id'``) is returned (e.g. ``'XXXX.A'``).
+
+    range : tuple(int, int), optional
+        If this parameter is specified, only PDB IDs in this range
+        are selected from all matching PDB IDs and returned
+        (pagination).
+        The range is zero-indexed and the stop value is exclusive.
+    sort_by : str or Sorting, optional
+        If specified, the returned PDB IDs are sorted by the values
+        of the given field name.
+        A complete list of the available fields is documented at
+        `<https://search.rcsb.org/structure-search-attributes.html>`_.
+        and
+        `<https://search.rcsb.org/chemical-search-attributes.html>`_.
+        If a string is given sorting is performed in descending order.
+        To choose the order, a :class:`Sorting` object needs to be
+        provided.
+    group_by : Grouping
+        If this parameter is set, the PDB IDs that meet the query
+        requirements, are grouped according to the given criterion.
+    return_groups : boolean, optional
+        Only has effect, if `group_by` is set.
+        By default the representative with the highest rank in each
+        group is returned.
+        The rank is determined by the `sort_by` parameter of
+        :class:`Grouping` provided in `group_by`.
+        If set to true, groups containing all structures belonging to
+        the group are returned instead.
+    content_types : iterable of {"experimental", "computational"}, optional
+        Specify whether experimental and computational structures should
+        be included.
+        At least one of them needs to be specified.
+        By default only experimental structures are included.
+        Note, that identifiers for computational structures cannot be
+        downloaded via :func:`biotite.database.rcsb.fetch()` as they
+        point to *AlphaFold DB* and *ModelArchive*.
+
+    Returns
+    -------
+    ids : list of str or dict (str -> list of str)
+        If `return_groups` is false (default case), a list of strings
+        containing all PDB IDs that meet the query requirements is
+        returned.
+        If `return_groups` is set to true a dictionary of groups is
+        returned.
+        This dictionary maps group identifiers to a list of all PDB IDs
+        belonging to this group.
+
+    Notes
+    -----
+    If `group_by` is set, the number of results may be lower than in an
+    ungrouped query, as grouping is not applicable to all structures.
+    For example a DNA structure has no associated *Uniprot* accession
+    and hence is omitted by :class:`UniprotGrouping`.
+
+    Also note that `sort_by` does not affect the order within a group.
+    This order is determined by the `sort_by` parameter of the
+    :class:`Grouping`.
+
+    Examples
+    --------
+
+    >>> query = FieldQuery("reflns.d_resolution_high", less_or_equal=0.6)
+    >>> print(sorted(search(query)))
+    ['1EJG', '1I0T', '3NIR', '3P4J', '4JLJ', '5D8V', '5NW3', '7ATG', '7R0H']
+    >>> print(search(query, sort_by="rcsb_accession_info.initial_release_date"))
+    ['7R0H', '7ATG', '5NW3', '5D8V', '4JLJ', '3P4J', '3NIR', '1I0T', '1EJG']
+    >>> print(search(
+    ...     query, range=(1,4), sort_by="rcsb_accession_info.initial_release_date"
+    ... ))
+    ['7ATG', '5NW3', '5D8V']
+    >>> print(sorted(search(query, return_type="polymer_instance")))
+    ['1EJG.A', '1I0T.A', '1I0T.B', '3NIR.A', '3P4J.A', '3P4J.B', '4JLJ.A', '4JLJ.B', '5D8V.A', '5NW3.A', '7ATG.A', '7ATG.B', '7R0H.A']
+    >>> print(search(
+    ...     query, return_type="polymer_entity", return_groups=True,
+    ...     group_by=UniprotGrouping(sort_by="rcsb_accession_info.initial_release_date"),
+    ... ))
+    {'P24297': ['5NW3_1'], 'P27707': ['4JLJ_1'], 'P80176': ['5D8V_1'], 'O29777': ['7R0H_1'], 'P01542': ['3NIR_1', '1EJG_1']}
+    """
+    query_dict = _initialize_query_dict(query, return_type, group_by, content_types)
+
+    if group_by is not None:
+        if return_groups:
+            query_dict["request_options"]["group_by_return_type"] = "groups"
+        else:
+            query_dict["request_options"]["group_by_return_type"] = "representatives"
+
+    if sort_by is not None:
+        if isinstance(sort_by, Sorting):
+            sorting = sort_by
+        else:
+            sorting = Sorting(sort_by)
+        query_dict["request_options"]["sort"] = [sorting.get_content()]
+
+    if range is None:
+        query_dict["request_options"]["return_all_hits"] = True
+    elif range[1] <= range[0]:
+        raise ValueError("Range stop must be greater than range start")
+    else:
+        query_dict["request_options"]["paginate"] = {
+            "start": int(range[0]),
+            "rows": int(range[1]) - int(range[0]),
+        }
+
+    r = requests.get(_search_url, params={"json": json.dumps(query_dict)})
+
+    if r.status_code == 200:
+        if group_by is None or not return_groups:
+            return [result["identifier"] for result in r.json()["result_set"]]
+        else:
+            return {
+                group["identifier"]: [
+                    result["identifier"] for result in group["result_set"]
+                ]
+                for group in r.json()["group_set"]
+            }
+    elif r.status_code == 204:
+        # Search did not return any results
+        return []
+    else:
+        try:
+            raise RequestError(f"Error {r.status_code}: {r.json()['message']}")
+        except json.decoder.JSONDecodeError:
+            # In case there an error response without message
+            raise RequestError(f"Error {r.status_code}")
+
+
+def _initialize_query_dict(query, return_type, group_by, content_types):
+    """
+    Initialize the request parameter dictionary with attributes that
+    `count()` and `search()` have in common.
+    """
+    if return_type not in [
+        "entry",
+        "polymer_instance",
+        "assembly",
+        "polymer_entity",
+        "non_polymer_entity",
+    ]:
+        raise ValueError(f"'{return_type}' is an invalid return type")
+
+    request_options = {}
+
+    if len(content_types) == 0:
+        raise ValueError("At least one content type must be specified")
+    for content_type in content_types:
+        if content_type not in ("experimental", "computational"):
+            raise ValueError(f"Unknown content type '{content_type}'")
+    request_options["results_content_type"] = content_types
+
+    if group_by is not None:
+        if not group_by.is_compatible_return_type(return_type):
+            raise ValueError(
+                f"Return type '{return_type}' is not compatible with the given Grouping"
+            )
+        request_options["group_by"] = group_by.get_content()
+
+    query_dict = {
+        "query": query.get_content(),
+        "return_type": return_type,
+        "request_options": request_options,
+    }
+    return query_dict
+
+
+def _to_isoformat(object):
+    """
+    Convert a datetime into the specifc ISO 8601 format required by the RCSB.
+    """
+    return object.strftime("%Y-%m-%dT%H:%M:%SZ")