add error msg when parsePDB fails. Edit other error msgs to make them…

… more meaningful

rhapsody/features/PolyPhen2.py

@@ -225,10 +225,17 @@ def parsePolyPhen2output(pph2_output):
     # discard invalid lines
     lines = [l for l in lines if l.strip() and l[0] != '#']
     if not lines:
-        msg = ("PolyPhen-2's output is empty. Please make sure that: \n"
-               "1) variants' format is correct "
-               '(\"UniprotID pos wt_aa mut_aa\") \n'
-               "2) query contains *human* variants \n")
+        msg = (
+            "PolyPhen-2's output is empty. Please check file 'pph2-log.txt' "
+            "in the output folder for error messages from PolyPhen-2. \n"
+            "Typical errors include: \n"
+            "1) query contains *non-human* variants \n"
+            "2) variants' format is incorrect (e.g. "
+            '"UniprotID pos wt_aa mut_aa") \n'
+            "3) wild-type amino acids are in the wrong position on the "
+            "sequence (please refer to Uniprot's canonical isoform) \n"
+            "4) Uniprot accession number is not recognized by PolyPhen-2. \n"
+        )
         raise RuntimeError(msg)
     # define a structured array
     pl_dtype = np.dtype([(col, 'U25') for col in pph2_columns])

rhapsody/features/Uniprot.py

@@ -33,10 +33,11 @@ def queryUniprot(*args, n_attempts=3, dt=1, **kwargs):
     attempt = 0
     while attempt < n_attempts:
         try:
-            print(f'attempt {attempt}')
             _ = openURL('http://www.uniprot.org/')
             break
         except:
+            LOGGER.info(
+                f'Attempt {attempt} to contact www.uniprot.org failed')
             attempt += 1
             time.sleep((attempt+1)*dt)
     else:
@@ -172,7 +173,12 @@ def alignCustomPDB(self, PDB, chain='all', title=None, recover=False):
         assert isinstance(title, str) or title is None
         # parse/import pdb and assign title
         if isinstance(PDB, str):
-            pdb = pd.parsePDB(PDB, subset='calpha')
+            try:
+                pdb = pd.parsePDB(PDB, subset='calpha')
+            except Exception as e:
+                msg = ('Invalid PDB structure: the file might be '
+                       f'corrupted or contain errors. Error message: {e}')
+                raise LOGGER.error(msg)
             if title is None:
                 title = os.path.basename(PDB.strip())
                 title = title.replace(' ', '_')

rhapsody/predict/core.py

@@ -383,7 +383,7 @@ def getUniqueSAVcoords(self):
             except Exception:
                 LOGGER.warn(
                     'Invalid Uniprot coordinates at line {}: {}'.format(
-                    i, SAV['unique SAV coords']))
+                        i, SAV['unique SAV coords']))
                 uSAVcoords[i] = tuple(['?', '?', '?', -999, '?', '?'])
         return uSAVcoords