remove class SingletonFamily

Class MolecularFamily can be used to represent singleton molecular family.

src/nplinker/class_info/chem_classes.py

@@ -555,7 +555,7 @@ def _read_cf_classes(self, mne_dir):
                 nr_nodes = line.pop(0)
                 # todo: make it easier to query classes of singleton families
                 # if singleton family, format like '-1_spectrum-id' like canopus results
-                # CG: Note that the singleton families id is "singleton-" + uuid.
+                # Note that the singleton families id is "singleton-" + spectrum-id.
                 if nr_nodes == "1":
                     component = f"-1_{cluster}"
                 class_info = []

src/nplinker/metabolomics/__init__.py

@@ -1,10 +1,9 @@
 import logging
 from .molecular_family import MolecularFamily
-from .singleton_family import SingletonFamily
 from .spectrum import GNPS_KEY
 from .spectrum import Spectrum
 
 
 logging.getLogger(__name__).addHandler(logging.NullHandler())
 
-__all__ = ["MolecularFamily", "SingletonFamily", "GNPS_KEY", "Spectrum"]
+__all__ = ["MolecularFamily", "GNPS_KEY", "Spectrum"]

src/nplinker/nplinker.py

@@ -427,8 +427,6 @@ def get_common_strains(
     ) -> dict[tuple[Spectrum | MolecularFamily, GCF], list[Strain]]:
         """Get common strains between given spectra/molecular families and GCFs.
 
-        Note that SingletonFamily objects are excluded from given molecular families.
-
         Args:
             met(Sequence[Spectrum] | Sequence[MolecularFamily]):
                 A list of Spectrum or MolecularFamily objects.

src/nplinker/scoring/linking/data_links.py

@@ -6,7 +6,6 @@
 from nplinker.genomics.gcf import GCF
 from nplinker.logconfig import LogConfig
 from nplinker.metabolomics import MolecularFamily
-from nplinker.metabolomics import SingletonFamily
 from nplinker.metabolomics import Spectrum
 from .utils import calc_correlation_matrix
 from .utils import isinstance_all
@@ -108,8 +107,6 @@ def get_common_strains(
     ) -> dict[tuple[Spectrum | MolecularFamily, GCF], list[Strain]]:
         """Get common strains between given spectra/molecular families and GCFs.
 
-        Note that SingletonFamily objects are excluded from given `spectra_or_mfs`.
-
         Args:
             spectra_or_mfs(Sequence[Spectrum | MolecularFamily]):
                 A list of Spectrum and/or MolecularFamily objects.
@@ -138,8 +135,6 @@ def get_common_strains(
         strain_ids = self.occurrence_gcf_strain.columns
         results = {}
         for obj in spectra_or_mfs:
-            if isinstance(obj, SingletonFamily):
-                continue
             for gcf in gcfs:
                 if isinstance(obj, Spectrum):
                     shared_strains = strain_ids[
@@ -215,12 +210,7 @@ def _get_occurrence_mf_strain(
         strains as columns, where index is `mf.family_id` and column name is
         `strain.id`. The values are 1 if the molecular family contains the
         strain and 0 otherwise.
-
-        Note that SingletonFamily objects are excluded from given `mfs`.
         """
-        # remove SingletonFamily objects
-        mfs = [mf for mf in mfs if not isinstance(mf, SingletonFamily)]
-
         df_mf_strain = pd.DataFrame(
             np.zeros((len(mfs), len(strains))),
             index=[mf.family_id for mf in mfs],

tests/scoring/test_data_links.py

@@ -1,7 +1,6 @@
 import pandas as pd
 import pytest
 from pandas.testing import assert_frame_equal
-from nplinker.metabolomics import SingletonFamily
 
 
 def test_init(datalinks):
@@ -173,36 +172,6 @@ def test_get_common_strains_spec_mf(datalinks, spectra, mfs, gcfs, strains_list)
     assert sut == expected
 
 
-def test_get_common_strains_sf(datalinks, mfs, gcfs, strains_list):
-    """Test get_common_strains method for input SingletonFamily."""
-    smf = SingletonFamily()
-
-    sut = datalinks.get_common_strains([smf], gcfs)
-    assert sut == {}
-
-    # the expected are same as `test_get_common_strains_mf`
-    mfs_mix = (*mfs[:2], smf)
-    sut = datalinks.get_common_strains(mfs_mix, gcfs)
-    expected = {
-        (mfs[0], gcfs[0]): [strains_list[0]],
-        (mfs[0], gcfs[1]): [],
-        (mfs[0], gcfs[2]): [strains_list[0]],
-        (mfs[1], gcfs[0]): [],
-        (mfs[1], gcfs[1]): [strains_list[1]],
-        (mfs[1], gcfs[2]): [strains_list[1]],
-    }
-    assert sut == expected
-
-    sut = datalinks.get_common_strains(mfs_mix, gcfs, filter_no_shared=True)
-    expected = {
-        (mfs[0], gcfs[0]): [strains_list[0]],
-        (mfs[0], gcfs[2]): [strains_list[0]],
-        (mfs[1], gcfs[1]): [strains_list[1]],
-        (mfs[1], gcfs[2]): [strains_list[1]],
-    }
-    assert sut == expected
-
-
 def test_get_common_strains_invalid_value(datalinks, spectra, gcfs):
     """Test get_common_strains method for empty arguments."""
     with pytest.raises(ValueError) as e: