diff --git a/jaqpotpy/datasets/molecular_datasets.py b/jaqpotpy/datasets/molecular_datasets.py
index 1f7afc40..49825c87 100644
--- a/jaqpotpy/datasets/molecular_datasets.py
+++ b/jaqpotpy/datasets/molecular_datasets.py
@@ -34,7 +34,10 @@ def __init__(self, df: pd.DataFrame = None, path: Optional[str] = None,
                  featurizer: Optional[MolecularFeaturizer] = None,
                  task:str = None
                  ) -> None:
-       
+        
+        if not(isinstance(y_cols, str) or (isinstance(y_cols, list) and isinstance(y_cols[1], str))):
+               raise ValueError("y_cols must be provided and should be either a string or a list of strings") 
+
         super().__init__(df=df, path=path, y_cols=y_cols, x_cols=x_cols, task = task)
         
         if isinstance(smiles_cols, str):
@@ -47,7 +50,7 @@ def __init__(self, df: pd.DataFrame = None, path: Optional[str] = None,
             self.smiles_cols = None
             self.smiles_cols_len = 0
         else:
-            raise TypeError("smiles_cols must either be a string, a list of strings or a None.")
+            raise TypeError("smiles_cols must either be a string, a list of strings or None.")
         
         self.featurizer = featurizer
         self._featurizer_name = None
@@ -74,16 +77,22 @@ def x_cols_all(self, value):
     def create(self):
 
         if (isinstance(self.smiles_cols,list) and len(self.smiles_cols) == 1):
+            # The method featurize_dataframe needs self.smiles to be pd.Series
             self.smiles = self._df[self.smiles_cols[0]]
             descriptors = self.featurizer.featurize_dataframe(self.smiles)
         elif isinstance(self.smiles_cols,str):  
             self.smiles = self._df[self.smiles_cols]
             descriptors = self.featurizer.featurize_dataframe(self.smiles) 
-        else:
+        elif:
             featurized_dfs = [self.featurizer.featurize_dataframe(self._df[[col]]) for col in self.smiles_cols]
             descriptors = pd.concat(featurized_dfs, axis=1)
-
+        else:
+            #Case where no smiles were provided
+            self.smiles = None
+            descriptors = []
+        
         self._y = self._df[self.y_cols]
+
         if self.x_cols is None:
             # Estimate x_cols by excluding y_cols and smiles_col
             x_ext =  pd.concat(self._df.drop(columns=self.y_cols + [self.smiles_cols]), descriptors)
@@ -97,25 +106,25 @@ def create(self):
         self._df = pd.concat([self._x, self._y], axis = 1)
 
     def __get_x__(self):
-        return self.df[self._x].to_numpy()
+        return self._df[self._x].to_numpy()
 
     def __get_y__(self):
-        return self.df[self._y].to_numpy()
+        return self._df[self._y].to_numpy()
 
     def __get__(self, instance, owner):
         if instance is None:
             return self
-        return instance.__dict__[self.df]
+        return instance.__dict__[self._df]
 
     def __getitem__(self, idx):
         # print(self.df[self.X].iloc[idx].values)
         # print(type(self.df[self.X].iloc[idx].values))
-        selected_x = self.df[self._x].iloc[idx].values
-        selected_y = self.df[self._y].iloc[idx].to_numpy()
+        selected_x = self._df[self._x].iloc[idx].values
+        selected_y = self._df[self._y].iloc[idx].to_numpy()
         return selected_x, selected_y
 
     def __len__(self):
-        return len(self.df)
+        return len(self._df)
 
     def __repr__(self) -> str:
         return (f"{self.__class__.__name__}"