add keep_mibig_only to BigscapeGCFLoader.get_gcfs method

- add parameter `keep_mibig_only`
- update and add unit tests

src/nplinker/genomics/bigscape/bigscape_loader.py

@@ -25,8 +25,18 @@ def __init__(self, cluster_file: str | PathLike, /) -> None:
         self._gcf_dict = self._parse_gcf(self.cluster_file)
         self._gcf_list = list(self._gcf_dict.values())
 
-    def get_gcfs(self) -> list[GCF]:
-        """Get all GCF objects."""
+    def get_gcfs(self, keep_mibig_only=False) -> list[GCF]:
+        """Get all GCF objects.
+
+        Args:
+            keep_mibig_only(bool): True to keep GCFs that contain only MIBiG
+                BGCs.
+
+        Returns:
+            list[GCF]: a list of GCF objects.
+        """
+        if not keep_mibig_only:
+            return [gcf for gcf in self._gcf_list if not gcf.has_mibig_only()]
         return self._gcf_list
 
     @staticmethod

tests/genomics/test_bigscape_loader.py

@@ -22,13 +22,19 @@ def test_init(self, loader):
                                           "mix_clustering_c0.30.tsv")
 
     def test_get_gcfs(self, loader):
-        gcfs = loader.get_gcfs()
+        gcfs = loader.get_gcfs(keep_mibig_only=True)
         assert isinstance(gcfs, list)
         assert len(gcfs) == 114
         assert isinstance(gcfs[0], GCF)
 
+    def test_get_gcfs_without_mibig_only(self, loader):
+        gcfs = loader.get_gcfs(keep_mibig_only=False)
+        assert isinstance(gcfs, list)
+        assert len(gcfs) == 113
+        assert isinstance(gcfs[0], GCF)
+
     def test_parse_gcf(self, loader):
-        gcf_dict = BigscapeGCFLoader._parse_gcf(loader.cluster_file) # noqa
+        gcf_dict = BigscapeGCFLoader._parse_gcf(loader.cluster_file)  # noqa
         assert isinstance(gcf_dict, dict)
         assert len(gcf_dict) == 114
         gcf = gcf_dict["135"]