Merge pull request #174 from AurelienJaquier/nwb-select

select nwb traces according to v_file field

bluepyefe/cell.py

@@ -63,7 +63,8 @@ def reader(self, config_data, recording_reader=None):
 
         if "v_file" in config_data:
             filename = config_data["v_file"]
-        elif "filepath" in config_data:
+        # if both present: use filepath. e.g. for some nwb that 'contain' igor files
+        if "filepath" in config_data:
             filename = config_data["filepath"]
 
         if recording_reader:

bluepyefe/nwbreader.py

@@ -5,18 +5,20 @@
 
 
 class NWBReader:
-    def __init__(self, content, target_protocols, repetition=None):
+    def __init__(self, content, target_protocols, repetition=None, v_file=None):
         """ Init
 
         Args:
             content (h5.File): NWB file
             target_protocols (list of str): list of the protocols to be read and returned
             repetition (list of int): id of the repetition(s) to be read and returned
+            v_file (str): name of original file that can be retrieved in sweep's description
         """
 
         self.content = content
         self.target_protocols = target_protocols
         self.repetition = repetition
+        self.v_file = v_file
 
     def read(self):
         """ Read the content of the NWB file
@@ -84,7 +86,10 @@ def read(self):
             if not isinstance(protocol_name, str):
                 protocol_name = protocol_name.decode('UTF-8')
 
-            if self.target_protocols and protocol_name not in self.target_protocols:
+            if (
+                self.target_protocols and
+                protocol_name.lower() not in [prot.lower() for prot in self.target_protocols]
+            ):
                 continue
 
             data.append(self._format_nwb_trace(
@@ -111,7 +116,10 @@ def read(self):
             key_current = sweep.replace('Series', 'StimulusSeries')
             protocol_name = "Step"
 
-            if self.target_protocols and protocol_name not in self.target_protocols:
+            if (
+                self.target_protocols and
+                protocol_name.lower() not in [prot.lower() for prot in self.target_protocols]
+            ):
                 continue
 
             if key_current not in self.content['stimulus']['presentation']:
@@ -162,8 +170,21 @@ def read(self):
         for ecode in self.target_protocols:
             for cell_id in self.content["data_organization"].keys():
                 if ecode not in self.content["data_organization"][cell_id]:
-                    logger.debug(f"No eCode {ecode} in nwb.")
-                    continue
+                    new_ecode = next(
+                        iter(
+                            ec
+                            for ec in self.content["data_organization"][cell_id]
+                            if ec.lower() == ecode.lower()
+                        )
+                    )
+                    if new_ecode:
+                        logger.debug(
+                            f"Could not find {ecode} in nwb file, will use {new_ecode} instead"
+                        )
+                        ecode = new_ecode
+                    else:
+                        logger.debug(f"No eCode {ecode} in nwb.")
+                        continue
 
                 ecode_content = self.content["data_organization"][cell_id][ecode]
 
@@ -185,11 +206,26 @@ def read(self):
                                 logger.debug(f"Ignoring {key_current} not"
                                              " present in the stimulus presentation")
                                 continue
+
                             if trace_name not in self.content["acquisition"]:
                                 logger.debug(f"Ignoring {trace_name} not"
                                              " present in the acquisition")
                                 continue
 
+                            # if we have v_file, check that trace comes from this original file
+                            if self.v_file is not None:
+                                attrs = self.content["acquisition"][trace_name].attrs
+                                if "description" not in attrs:
+                                    logger.warning(
+                                        "Ignoring %s because no description could be found.",
+                                        trace_name
+                                    )
+                                    continue
+                                v_file_end = self.v_file.split("/")[-1]
+                                if v_file_end != attrs.get("description", "").split("/")[-1]:
+                                    logger.debug(f"Ignoring {trace_name} not matching v_file")
+                                    continue
+
                             data.append(self._format_nwb_trace(
                                 voltage=self.content["acquisition"][trace_name]["data"],
                                 current=self.content["stimulus"]["presentation"][key_current][

bluepyefe/reader.py

@@ -191,7 +191,12 @@ def nwb_reader(in_data):
 
     with h5py.File(in_data["filepath"], "r") as content:
         if "data_organization" in content:
-            reader = BBPNWBReader(content, target_protocols, in_data.get("repetition", None))
+            reader = BBPNWBReader(
+                content,
+                target_protocols,
+                in_data.get("repetition", None),
+                in_data.get("v_file", None)
+            )
         elif "timeseries" in content["acquisition"].keys():
             reader = AIBSNWBReader(content, target_protocols)
         else: