bugfix in confidence, data must be loaded from unshuffled train.h5

package/chemicalchecker/core/sign3.py

@@ -121,7 +121,7 @@ class sign3(BaseSignature, DataSignature):
                 return None
             full_trim = np.argwhere(np.repeat(self.trim_mask, 128))
             self._sharedx_trim = self.sharedx[:, full_trim.ravel()]
-        self.__log.debug("sharedx_trim shape: %s" % 
+        self.__log.debug("sharedx_trim shape: %s" %
                          str(self._sharedx_trim.shape))
         return self._sharedx_trim
 
@@ -249,7 +249,7 @@ class sign3(BaseSignature, DataSignature):
         if ref_cc is not None:
             cc_ref = ChemicalChecker(ref_cc)
         else:
-            cc_ref = ChemicalChecker()
+            cc_ref = self.get_cc()
         sign3.__log.info(
             "Reference CC (for predict methods): %s" % cc_ref.cc_root)
         # create CC instance to host signatures
@@ -470,7 +470,8 @@ class sign3(BaseSignature, DataSignature):
         return siamese, prior_model, prior_sign_model, confidence_model
 
     def train_confidence(self, siamese, suffix='eval', traintest_file=None,
-                         max_x=10000, max_neig=50000, p_self=0.0):
+                         train_file=None, max_x=10000, max_neig=50000,
+                         p_self=0.0):
         """Train confidence and prior models."""
         # get sorted keys from siamese traintest file
         self.update_status('Training applicability')
@@ -479,25 +480,30 @@ class sign3(BaseSignature, DataSignature):
                                           'traintest_%s.h5' % suffix)
         if not os.path.isfile(traintest_file):
             raise Exception('Traintest_file not found: %s' % traintest_file)
+        if train_file is None:
+            train_file = os.path.join(self.model_path, 'train.h5')
+        if not os.path.isfile(train_file):
+            raise Exception('Train_file not found: %s' % train_file)
         self.traintest_file = traintest_file
-        tt = DataSignature(self.traintest_file)
-        test_inks = tt.get_h5_dataset('keys_test')[:max_x]
-        test_inks = np.sort(test_inks)
-        train_inks = tt.get_h5_dataset('keys_train')[:max_neig]
-        train_inks = np.sort(train_inks)
+        traintest = DataSignature(self.traintest_file)
+        test_inks = traintest.get_h5_dataset('keys_test')[:max_x]
+        train_inks = traintest.get_h5_dataset('keys_train')[:max_neig]
+
+        # confidence is going to be trained only on siamese test data
         test_mask = np.isin(list(self.sign2_self.keys), list(test_inks),
-                            assume_unique=True)
+                                  assume_unique=True)
         train_mask = np.isin(list(self.sign2_self.keys), list(train_inks),
-                             assume_unique=True)
-        # confidence is going to be trained only on siamese test data
-        confidence_train_x = self.sharedx[test_mask]
-        s2_test = self.sign2_self.get_h5_dataset('V', mask=test_mask)
+                                   assume_unique=True)
+        train = DataSignature(train_file)
+        confidence_train_x = train.get_h5_dataset('x', mask=test_mask)
+        #s2_test = self.sign2_self.get_h5_dataset('V', mask=test_mask)
+        _, s2_test = self.sign2_self.get_vectors(test_inks)
         s2_test_x = confidence_train_x[:, self.dataset_idx[0]
                                        * 128: (self.dataset_idx[0] + 1) * 128]
-        self.__log.debug('self.dataset_idx: %s' % str(self.dataset_idx))
+
         assert(np.all(s2_test == s2_test_x))
         # siamese train is going to be used for appticability domain
-        known_x = self.sharedx[train_mask]
+        known_x = train.get_h5_dataset('x', mask=train_mask)
         # generate train-test split for confidence estimation
         split_names = ['train', 'test']
         split_fractions = [0.8, 0.2]
@@ -2238,7 +2244,7 @@ def plot_subsample(sign, plot_file, sign2_coverage, traintest_file,
     import matplotlib.pyplot as plt
     from chemicalchecker import ChemicalChecker
 
-    cc = ChemicalChecker()
+    cc = sign.get_cc()
 
     # NICO sign2_list
     if sign2_list is not None: