Handle missing structural data more nicely.

alignementseq_tests.py

@@ -28,7 +28,9 @@ class BalibaseTestCase(unittest.TestCase):
     def setUp(self):
         balibase_zippath = "balibase.zip"
         self.balibase_path = "balibase"
-        self.exclude_fastas = ["BBS11037.fasta"] # This one does not contain correct
+        self.exclude_fastas = [] # For problematic files that won't be handled.
+                                 # Try to handle them, this is a temporary or
+                                 # last resort solution.
         # pdb codes as ids
         testfile_path = os.path.join(
             self.balibase_path, 
@@ -88,18 +90,35 @@ class BalibaseTestCase(unittest.TestCase):
                 warnings.warn("Exluding file "+name+" according to exclusion list")
                 continue
 
+            skip = False
             item = [s1, s2, os1, os2]
+
             for s in s1, s2:
-                fname = Bio.PDB.PDBList().retrieve_pdb_file(
-                        s.id[0:4],
-                        file_format="mmCif",
-                        pdir="data"
-                        )
+                code = s.id[0:4]
+                fname = os.path.join('data', code+'.cif')
+
+                # Attempt download
+                if not os.path.isfile(fname):
+                    fname = Bio.PDB.PDBList().retrieve_pdb_file(
+                            code,
+                            file_format="mmCif",
+                            pdir="data"
+                            )
+
+                # Skip if still missing
+                if not os.path.isfile(fname):
+                    warnings.warn("Could not retrieve PDB "+code+" (from "+name+"), skipping")
+                    skip = True
+                    break
+
                 with warnings.catch_warnings():
                     warnings.simplefilter("ignore")
-                    struct = parser.get_structure(s.id[0:4], fname)
-                print(fname, s.id[0:4])
+                    struct = parser.get_structure(code, fname)
                 item.append(struct)
+
+            if skip:
+                continue
+
             item.append(name)
             yield item
 
@@ -238,7 +257,7 @@ class AlignmentSeqTestCase(BalibaseTestCase):
                             str(Bio.PDB.Polypeptide.Polypeptide(c).get_sequence()).startswith(str(seq.seq))
                             )
                 except StopIteration:
-                    warnings.warn("No suitable chain found for seq id "+seq.id+" in structure "+struct.id+", skipping file.")
+                    warnings.warn("No suitable chain found for seq id "+seq.id+" in structure "+struct.id+" (from "+name+"), skipping file.")
                     skip = True
                     break
                 polyp = Bio.PDB.Polypeptide.Polypeptide(chain)