Fix guess_TopologyAttrs for empty Elements and Atomtypes

package/MDAnalysis/core/topologyattrs.py

@@ -247,18 +247,14 @@ def stub_method(self, *args, **kwargs):
         )
         raise NoDataError(message)
 
-    annotation = textwrap.dedent(
-        """\
+    annotation = textwrap.dedent("""\
         .. note::
 
           This requires the underlying topology to have {}. Otherwise, a
           :exc:`~MDAnalysis.exceptions.NoDataError` is raised.
 
 
-    """.format(
-            attribute_name
-        )
-    )
+    """.format(attribute_name))
     # The original docstring is assumed to be formatted with black
     # (i.e., *all text* lines are already indented, the first line
     # is a triple quote followed by newline)
@@ -396,9 +392,7 @@ def __init__(cls, name, bases, classdict):
             if dtype is not None:
                 per_obj = classdict.get("per_object", bases[0].per_object)
                 try:
-                    selection.gen_selection_class(
-                        singular, attrname, dtype, per_obj
-                    )
+                    selection.gen_selection_class(singular, attrname, dtype, per_obj)
                 except ValueError:
                     msg = (
                         "A selection keyword could not be "
@@ -476,9 +470,7 @@ def _gen_initial_values(n_atoms, n_residues, n_segments):
         raise NotImplementedError("No default values")
 
     @classmethod
-    def from_blank(
-        cls, n_atoms=None, n_residues=None, n_segments=None, values=None
-    ):
+    def from_blank(cls, n_atoms=None, n_residues=None, n_segments=None, values=None):
         """Create a blank version of this TopologyAttribute
 
         Parameters
@@ -829,9 +821,7 @@ def _set_X(self, ag, values):
                     newnames.append(val)
                     newidx[i] = nextidx
 
-        self.nmidx[ag.ix] = (
-            newidx  # newidx either single value or same size array
-        )
+        self.nmidx[ag.ix] = newidx  # newidx either single value or same size array
         if newnames:
             self.name_lookup = np.concatenate([self.name_lookup, newnames])
         self.values = self.name_lookup[self.nmidx]
@@ -954,8 +944,7 @@ def phi_selections(residues, c_name="C", n_name="N", ca_name="CA"):
 
         keep_prev = [sum(r.atoms.names == c_name) == 1 for r in prev]
         keep_res = [
-            all(sum(r.atoms.names == n) == 1 for n in ncac_names)
-            for r in residues
+            all(sum(r.atoms.names == n) == 1 for n in ncac_names) for r in residues
         ]
         keep = np.array(keep_prev) & np.array(keep_res)
         keep[invalid] = False
@@ -1146,8 +1135,7 @@ def psi_selections(residues, c_name="C", n_name="N", ca_name="CA"):
 
         keep_nxt = [sum(r.atoms.names == n_name) == 1 for r in nxt]
         keep_res = [
-            all(sum(r.atoms.names == n) == 1 for n in ncac_names)
-            for r in residues
+            all(sum(r.atoms.names == n) == 1 for n in ncac_names) for r in residues
         ]
         keep = np.array(keep_nxt) & np.array(keep_res)
         nxt = nxt[keep]
@@ -1259,12 +1247,9 @@ def omega_selections(residues, c_name="C", n_name="N", ca_name="CA"):
 
         nxtatoms = [ca_name, n_name]
         resatoms = [ca_name, c_name]
-        keep_nxt = [
-            all(sum(r.atoms.names == n) == 1 for n in nxtatoms) for r in nxt
-        ]
+        keep_nxt = [all(sum(r.atoms.names == n) == 1 for n in nxtatoms) for r in nxt]
         keep_res = [
-            all(sum(r.atoms.names == n) == 1 for n in resatoms)
-            for r in residues
+            all(sum(r.atoms.names == n) == 1 for n in resatoms) for r in residues
         ]
         keep = np.array(keep_nxt) & np.array(keep_res)
         nxt = nxt[keep]
@@ -1390,6 +1375,7 @@ class Atomtypes(AtomStringAttr):
     singular = "type"
     per_object = "atom"
     dtype = object
+    missing_value_label = ""
 
 
 # TODO: update docs to property doc
@@ -1399,6 +1385,7 @@ class Elements(AtomStringAttr):
     attrname = "elements"
     singular = "element"
     dtype = object
+    missing_value_label = ""
 
     @staticmethod
     def _gen_initial_values(na, nr, ns):
@@ -1766,13 +1753,9 @@ def moment_of_inertia(group, wrap=False, unwrap=False, compound="group"):
         """
         atomgroup = group.atoms
 
-        com = atomgroup.center_of_mass(
-            wrap=wrap, unwrap=unwrap, compound=compound
-        )
+        com = atomgroup.center_of_mass(wrap=wrap, unwrap=unwrap, compound=compound)
         if compound != "group":
-            com = (com * group.masses[:, None]).sum(
-                axis=0
-            ) / group.masses.sum()
+            com = (com * group.masses[:, None]).sum(axis=0) / group.masses.sum()
 
         if wrap:
             pos = atomgroup.pack_into_box(inplace=False) - com
@@ -1903,9 +1886,7 @@ def _gyration(recenteredpos, masses):
         atomgroup = group.atoms
         masses = atomgroup.masses
 
-        com = atomgroup.center_of_mass(
-            wrap=wrap, unwrap=unwrap, compound=compound
-        )
+        com = atomgroup.center_of_mass(wrap=wrap, unwrap=unwrap, compound=compound)
 
         if compound == "group":
             if wrap:
@@ -1923,7 +1904,7 @@ def _gyration(recenteredpos, masses):
                 recenteredpos = atomgroup.positions - com
             eig_vals = _gyration(recenteredpos, masses)
         else:
-            (atom_masks, compound_masks, n_compounds) = (
+            atom_masks, compound_masks, n_compounds = (
                 atomgroup._split_by_compound_indices(compound)
             )
 
@@ -2040,8 +2021,7 @@ def asphericity(group, wrap=False, unwrap=False, compound="group"):
             )
         else:
             shape = (3.0 / 2.0) * (
-                np.sum((eig_vals - np.mean(eig_vals)) ** 2)
-                / np.sum(eig_vals) ** 2
+                np.sum((eig_vals - np.mean(eig_vals)) ** 2) / np.sum(eig_vals) ** 2
             )
 
         return shape
@@ -2125,9 +2105,7 @@ def align_principal_axis(group, axis, vector):
         # print "axis = %r, angle = %f deg" % (ax, angle)
         return group.rotateby(angle, ax)
 
-    transplants[GroupBase].append(
-        ("align_principal_axis", align_principal_axis)
-    )
+    transplants[GroupBase].append(("align_principal_axis", align_principal_axis))
 
 
 # TODO: update docs to property doc
@@ -2284,9 +2262,7 @@ def total_charge(group, compound="group"):
     @warn_if_not_unique
     @_pbc_to_wrap
     @check_wrap_and_unwrap
-    def dipole_vector(
-        group, wrap=False, unwrap=False, compound="group", center="mass"
-    ):
+    def dipole_vector(group, wrap=False, unwrap=False, compound="group", center="mass"):
         r"""Dipole vector of the group.
 
         .. math::
@@ -2352,9 +2328,7 @@ def dipole_vector(
 
         if center == "mass":
             masses = atomgroup.masses
-            ref = atomgroup.center_of_mass(
-                wrap=wrap, unwrap=unwrap, compound=compound
-            )
+            ref = atomgroup.center_of_mass(wrap=wrap, unwrap=unwrap, compound=compound)
         elif center == "charge":
             ref = atomgroup.center_of_charge(
                 wrap=wrap, unwrap=unwrap, compound=compound
@@ -2380,11 +2354,9 @@ def dipole_vector(
                 )
             else:
                 recenteredpos = atomgroup.positions - ref
-            dipole_vector = np.einsum(
-                "ij,ij->j", recenteredpos, charges[:, np.newaxis]
-            )
+            dipole_vector = np.einsum("ij,ij->j", recenteredpos, charges[:, np.newaxis])
         else:
-            (atom_masks, compound_masks, n_compounds) = (
+            atom_masks, compound_masks, n_compounds = (
                 atomgroup._split_by_compound_indices(compound)
             )
 
@@ -2466,13 +2438,9 @@ def dipole_moment(group, **kwargs):
         dipole_vector = atomgroup.dipole_vector(**kwargs)
 
         if len(dipole_vector.shape) > 1:
-            dipole_moment = np.sqrt(
-                np.einsum("ij,ij->i", dipole_vector, dipole_vector)
-            )
+            dipole_moment = np.sqrt(np.einsum("ij,ij->i", dipole_vector, dipole_vector))
         else:
-            dipole_moment = np.sqrt(
-                np.einsum("i,i->", dipole_vector, dipole_vector)
-            )
+            dipole_moment = np.sqrt(np.einsum("i,i->", dipole_vector, dipole_vector))
 
         return dipole_moment
 
@@ -2565,9 +2533,7 @@ def __quadrupole_tensor(recenteredpos, charges):
 
         if center == "mass":
             masses = atomgroup.masses
-            ref = atomgroup.center_of_mass(
-                wrap=wrap, unwrap=unwrap, compound=compound
-            )
+            ref = atomgroup.center_of_mass(wrap=wrap, unwrap=unwrap, compound=compound)
         elif center == "charge":
             ref = atomgroup.center_of_charge(
                 wrap=wrap, unwrap=unwrap, compound=compound
@@ -2595,7 +2561,7 @@ def __quadrupole_tensor(recenteredpos, charges):
                 recenteredpos = atomgroup.positions - ref
             quad_tensor = __quadrupole_tensor(recenteredpos, charges)
         else:
-            (atom_masks, compound_masks, n_compounds) = (
+            atom_masks, compound_masks, n_compounds = (
                 atomgroup._split_by_compound_indices(compound)
             )
 
@@ -2685,9 +2651,7 @@ def __quadrupole_moment(tensor):
         if len(quad_tensor.shape) == 2:
             quad_moment = __quadrupole_moment(quad_tensor)
         else:
-            quad_moment = np.array(
-                [__quadrupole_moment(x) for x in quad_tensor]
-            )
+            quad_moment = np.array([__quadrupole_moment(x) for x in quad_tensor])
 
         return quad_moment
 
@@ -3011,9 +2975,7 @@ def sequence(self, **kwargs):
                 )
             )
         try:
-            sequence = "".join(
-                [convert_aa_code(r) for r in self.residues.resnames]
-            )
+            sequence = "".join([convert_aa_code(r) for r in self.residues.resnames])
         except KeyError as err:
             errmsg = (
                 f"AtomGroup contains a residue name '{err.message}' that"
@@ -3146,14 +3108,13 @@ def _check_connection_values(func):
     @functools.wraps(func)
     def wrapper(self, values, *args, **kwargs):
         if not all(
-            len(x) == self._n_atoms
-            and all(isinstance(y, (int, np.integer)) for y in x)
+            len(x) == self._n_atoms and all(isinstance(y, (int, np.integer)) for y in x)
             for x in values
         ):
             raise ValueError(
-                (
-                    "{} must be an iterable of tuples with {}" " atom indices"
-                ).format(self.attrname, self._n_atoms)
+                ("{} must be an iterable of tuples with {}" " atom indices").format(
+                    self.attrname, self._n_atoms
+                )
             )
         clean = []
         for v in values:
@@ -3231,9 +3192,7 @@ def _bondDict(self):
         """Lazily built mapping of atoms:bonds"""
         bd = defaultdict(list)
 
-        for b, t, g, o in zip(
-            self.values, self.types, self._guessed, self.order
-        ):
+        for b, t, g, o in zip(self.values, self.types, self._guessed, self.order):
             for a in b:
                 bd[a].append((b, t, g, o))
         return bd
@@ -3252,9 +3211,7 @@ def get_atoms(self, ag):
 
         """
         try:
-            unique_bonds = set(
-                itertools.chain(*[self._bondDict[a] for a in ag.ix])
-            )
+            unique_bonds = set(itertools.chain(*[self._bondDict[a] for a in ag.ix]))
         except TypeError:
             # maybe we got passed an Atom
             unique_bonds = self._bondDict[ag.ix]

testsuite/MDAnalysisTests/guesser/test_base.py

@@ -26,7 +26,7 @@
 import pytest
 from MDAnalysis import _GUESSERS, _TOPOLOGY_ATTRS
 from MDAnalysis.core.topology import Topology
-from MDAnalysis.core.topologyattrs import Atomnames, Atomtypes, Masses
+from MDAnalysis.core.topologyattrs import Atomnames, Atomtypes, Elements, Masses
 from MDAnalysis.exceptions import NoDataError
 from MDAnalysis.guesser.base import GuesserBase, get_guesser
 from numpy.testing import assert_allclose, assert_equal
@@ -58,16 +58,13 @@ class TestGuesser1(GuesserBase):
     def test_guess_invalid_attribute(self):
         with pytest.raises(
             ValueError,
-            match="default guesser can not guess "
-            "the following attribute: foo",
+            match="default guesser can not guess " "the following attribute: foo",
         ):
             mda.Universe(datafiles.PDB_xsmall, to_guess=["foo"])
 
     def test_guess_attribute_with_missing_parent_attr(self):
         names = Atomnames(np.array(["C", "HB", "HA", "O"], dtype=object))
-        masses = Masses(
-            np.array([np.nan, np.nan, np.nan, np.nan], dtype=np.float64)
-        )
+        masses = Masses(np.array([np.nan, np.nan, np.nan, np.nan], dtype=np.float64))
         top = Topology(4, 1, 1, attrs=[names, masses])
         u = mda.Universe(top, to_guess=["masses"])
         assert_allclose(
@@ -88,9 +85,7 @@ def test_partial_guessing(self):
         masses = Masses(np.array([0, np.nan, np.nan, 0], dtype=np.float64))
         top = Topology(4, 1, 1, attrs=[types, masses])
         u = mda.Universe(top, to_guess=["masses"])
-        assert_allclose(
-            u.atoms.masses, np.array([0, 1.00800, 1.00800, 0]), atol=0
-        )
+        assert_allclose(u.atoms.masses, np.array([0, 1.00800, 1.00800, 0]), atol=0)
 
     def test_force_guess_priority(self):
         "check that passing the attribute to force_guess have higher power"
@@ -104,14 +99,21 @@ def test_force_guess_priority(self):
             atol=0,
         )
 
-    def test_partial_guess_attr_with_unknown_no_value_label(self):
-        "trying to partially guess attribute tha doesn't have declared"
-        "no_value_label should gives no effect"
+    def test_partial_guess_types_with_missing_value_label(self):
+        """Atomtypes with empty string values are guessed because
+        missing_value_label is defined as an empty string."""
         names = Atomnames(np.array(["C", "H", "H", "O"], dtype=object))
         types = Atomtypes(np.array(["", "", "", ""], dtype=object))
         top = Topology(4, 1, 1, attrs=[names, types])
         u = mda.Universe(top, to_guess=["types"])
-        assert_equal(u.atoms.types, ["", "", "", ""])
+        assert_equal(u.atoms.types, ["C", "H", "H", "O"])
+
+    def test_partial_guess_elements_with_missing_value_label(self):
+        names = Atomnames(np.array(["C1", "N1"], dtype=object))
+        elements = Elements(np.array(["C", ""], dtype=object))
+        top = Topology(2, 1, 1, attrs=[names, elements])
+        u = mda.Universe(top, to_guess=["elements"])
+        assert_equal(u.atoms.elements, ["C", "N"])
 
     def test_guess_topology_objects_existing_read(self):
         u = mda.Universe(datafiles.CONECT)
@@ -185,9 +187,7 @@ def test_guess_topology_objects_out_of_order_guess(self):
         with pytest.raises(NoDataError):
             u.atoms.angles
 
-        u.guess_TopologyAttrs(
-            "default", to_guess=["dihedrals", "angles", "bonds"]
-        )
+        u.guess_TopologyAttrs("default", to_guess=["dihedrals", "angles", "bonds"])
         assert len(u.atoms.angles) == 290
         assert len(u.atoms.dihedrals) == 411
 
@@ -249,9 +249,7 @@ def test_guess_singular(self):
 
 
 def test_Universe_guess_bonds_deprecated():
-    with pytest.warns(
-        DeprecationWarning, match="`guess_bonds` keyword is deprecated"
-    ):
+    with pytest.warns(DeprecationWarning, match="`guess_bonds` keyword is deprecated"):
         u = mda.Universe(datafiles.PDB_xsmall, guess_bonds=True)