Library temporary fix (by markmipt)

Author: RobbinBouwmeester

CHANGELOG.md

@@ -5,6 +5,9 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to
 [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.1.32] - 2021-07-29
+- Temporary fix suggested by markmipt for the library
+
 ## [0.1.31] - 2021-07-01
 - Change logging to specified logger object instead of standard logger
 

deeplc/deeplc.py

@@ -142,7 +142,8 @@ class DeepLC():
         Make predictions
 
     """
-
+    library = {}
+    
     def __init__(self,
                  main_path=os.path.dirname(os.path.realpath(__file__)),
                  path_model=None,
@@ -189,7 +190,6 @@ def __init__(self,
 
         self.use_library = use_library
         self.write_library = write_library
-        self.library = {}
 
         if self.use_library:
             self.read_library()
@@ -243,7 +243,7 @@ def read_library(self):
         for line_num,line in enumerate(library_file):
             split_line = line.strip().split(",")
             try:
-                self.library[split_line[0]] = float(split_line[1])
+                DeepLC.library[split_line[0]] = float(split_line[1])
             except:
                 logger.warning(
                     "Could not use this library entry due to an error: %s", line
@@ -433,7 +433,7 @@ def make_preds_core(self,
             all_mods = [m_name for m_group_name,m_name in self.model.items()]
 
         # TODO check if .keys() object is the same as set (or at least for set operations)
-        idents_in_lib = set(self.library.keys())
+        idents_in_lib = set(DeepLC.library.keys())
 
         if self.use_library:
             for ident in seq_df["idents"]:
@@ -479,6 +479,7 @@ def make_preds_core(self,
             cnn_verbose = 0
 
         # If we need to apply deep NN
+        del self.library
         if len(seq_df.index) > 0:
             if self.cnn_model:
                 if self.verbose:
@@ -504,6 +505,10 @@ def make_preds_core(self,
 
         ret_preds = []
         ret_preds2 = []
+        
+        self.library = {}
+        if self.use_library:
+            self.read_library()
 
         # If we need to calibrate
         if calibrate:
@@ -554,7 +559,7 @@ def make_preds_core(self,
                         p = list(self.calibration_core(uncal_preds,self.calibrate_dict[m_name],self.calibrate_min[m_name],self.calibrate_max[m_name]))
                         ret_preds.append(p)
 
-                        p2 = list(self.calibration_core([self.library[ri+"|"+m_name] for ri  in rem_idents],self.calibrate_dict[m_name],self.calibrate_min[m_name],self.calibrate_max[m_name]))
+                        p2 = list(self.calibration_core([DeepLC.library[ri+"|"+m_name] for ri  in rem_idents],self.calibrate_dict[m_name],self.calibrate_min[m_name],self.calibrate_max[m_name]))
                         ret_preds2.append(p2)
 
                     ret_preds = np.array([sum(a)/len(a) for a in zip(*ret_preds)])
@@ -597,7 +602,7 @@ def make_preds_core(self,
 
                     ret_preds = self.calibration_core(uncal_preds,self.calibrate_dict,self.calibrate_min,self.calibrate_max)
 
-                    p2 = list(self.calibration_core([self.library[ri+"|"+mod_name] for ri  in rem_idents],self.calibrate_dict,self.calibrate_min,self.calibrate_max))
+                    p2 = list(self.calibration_core([DeepLC.library[ri+"|"+mod_name] for ri  in rem_idents],self.calibrate_dict,self.calibrate_min,self.calibrate_max))
                     ret_preds2.extend(p2)
             else:
                 # first get uncalibrated prediction
@@ -649,7 +654,7 @@ def make_preds_core(self,
                                 lib_file.close()
                                 if self.reload_library: self.read_library()
 
-                            p2 = [self.library[ri+"|"+m_name] for ri  in rem_idents]
+                            p2 = [DeepLC.library[ri+"|"+m_name] for ri  in rem_idents]
                             ret_preds2.append(p2)
 
                         ret_preds = np.array([sum(a)/len(a) for a in zip(*ret_preds)])
@@ -677,7 +682,7 @@ def make_preds_core(self,
                             lib_file.close()
                             if self.reload_library: self.read_library()
 
-                        ret_preds2 = np.array([self.library[ri+"|"+mod_name] for ri  in rem_idents])
+                        ret_preds2 = np.array([DeepLC.library[ri+"|"+mod_name] for ri  in rem_idents])
                     elif isinstance(self.model, str):
                         # No library write!
                         mod_name = self.model
@@ -703,7 +708,7 @@ def make_preds_core(self,
                             lib_file.close()
                             if self.reload_library: self.read_library()
 
-                        ret_preds2 = np.array([self.library[ri+"|"+mod_name] for ri  in rem_idents])
+                        ret_preds2 = np.array([DeepLC.library[ri+"|"+mod_name] for ri  in rem_idents])
                     else:
                         raise DeepLCError('No CNN model defined.')
                 else:
@@ -732,7 +737,7 @@ def make_preds_core(self,
                             if self.reload_library: self.read_library()
                     except:
                         pass
-                    ret_preds2 = [self.library[ri+"|"+mod_name] for ri  in rem_idents]
+                    ret_preds2 = [DeepLC.library[ri+"|"+mod_name] for ri  in rem_idents]
 
             else:
                 # No library write!

…cp_cb975cfdd4105f97efa0b3afffe075cc.hdf5 …cp_cb975cfdd4105f97efa0b3afffe075cc.hdf5deeplc/models/full_hc_PXD005573_mcp_cb975cfdd4105f97efa0b3afffe075cc.hdf5 renamed to deeplc/mods/full_hc_PXD005573_mcp_cb975cfdd4105f97efa0b3afffe075cc.hdf5 deeplc/models/full_hc_PXD005573_mcp_cb975cfdd4105f97efa0b3afffe075cc.hdf5 renamed to deeplc/mods/full_hc_PXD005573_mcp_cb975cfdd4105f97efa0b3afffe075cc.hdf5

@@ -0,0 +1,42 @@
+"Unit and integration tests for DeepLC."
+
+# Standard library
+import logging
+import pytest
+import subprocess
+
+# Third party
+import pandas as pd
+from sklearn.metrics import r2_score
+
+# DeepLC
+import deeplc
+
+
+def test_cli_basic():
+    """ Test command line interface help message. """
+    assert subprocess.getstatusoutput('deeplc -h')[0] == 0, "`deeplc -h` \
+returned non-zero status code"
+
+
+def test_cli_full():
+    """" Test command line interface with input files."""
+    file_path_pred = "examples/datasets/test_train.csv"
+    file_path_cal = "examples/datasets/test_train.csv"
+    file_path_out = "pytest_cli_out.csv"
+
+    command = [
+        "deeplc", "--file_pred", file_path_pred, "--file_cal", file_path_cal,
+        "--file_pred_out", file_path_out
+    ]
+    subprocess.run(command, check=True)
+
+    preds_df = pd.read_csv(file_path_out)
+    model_r2 = r2_score(preds_df['tr'], preds_df['predicted_tr'])
+    logging.info("DeepLC R2 score on %s: %f", file_path_pred, model_r2)
+    assert model_r2 > 0.95, f"DeepLC R2 score on {file_path_pred} below 0.95 \
+(was {model_r2})"
+
+
+if __name__ == "__main__":
+    pytest.main()