Added clear_cache, do_CAR, invert_sign to GUI

kilosort/gui/main.py

@@ -340,6 +340,9 @@ def set_parameters(self):
 
         params = settings.copy()
         params['save_preprocessed_copy'] = self.run_box.save_preproc_check.isChecked()
+        params['clear_cache'] = self.run_box.clear_cache_check.isChecked()
+        params['do_CAR'] = self.run_box.do_CAR_check.isChecked()
+        params['invert_sign'] = self.run_box.invert_sign_check.isChecked()
 
         assert params
 

kilosort/gui/run_box.py

@@ -24,6 +24,9 @@ def __init__(self, parent):
         self.run_all_button = QtWidgets.QPushButton("Run")
         self.spike_sort_button = QtWidgets.QPushButton("Spikesort")
         self.save_preproc_check = QtWidgets.QCheckBox("Save Preprocessed Copy")
+        self.clear_cache_check = QtWidgets.QCheckBox("Clear PyTorch Cache")
+        self.do_CAR_check = QtWidgets.QCheckBox("CAR")
+        self.invert_sign_check = QtWidgets.QCheckBox("Invert Sign")
 
         self.buttons = [
             self.run_all_button
@@ -44,7 +47,7 @@ def __init__(self, parent):
         self.remote_widgets = None
 
         self.progress_bar = QtWidgets.QProgressBar()
-        self.layout.addWidget(self.progress_bar, 3, 0, 2, 2)
+        self.layout.addWidget(self.progress_bar, 5, 0, 3, 4)
 
         self.setup()
 
@@ -64,8 +67,36 @@ def setup(self):
             """
         self.save_preproc_check.setToolTip(preproc_text)
 
-        self.layout.addWidget(self.run_all_button, 0, 0, 2, 2)
-        self.layout.addWidget(self.save_preproc_check, 2, 0, 1, 2)
+        self.clear_cache_check.setCheckState(QtCore.Qt.CheckState.Unchecked)
+        cache_text = """
+            If enabled, force pytorch to free up memory reserved for its cache in
+            between memory-intensive operations.
+            Note that setting `clear_cache=True` is NOT recommended unless you
+            encounter GPU out-of-memory errors, since this can result in slower
+            sorting.
+            """
+        self.clear_cache_check.setToolTip(cache_text)
+
+        self.do_CAR_check.setCheckState(QtCore.Qt.CheckState.Checked)
+        car_text = """
+            If enabled, apply common average reference during preprocessing
+            (recommended).
+            """
+        self.do_CAR_check.setToolTip(car_text)
+
+        self.invert_sign_check.setCheckState(QtCore.Qt.CheckState.Unchecked)
+        invert_sign_text = """
+            If enabled, flip positive/negative values in data to conform to
+            standard expected by Kilosort4. This is NOT recommended unless you
+            know your data is using the opposite sign.
+            """
+        self.invert_sign_check.setToolTip(invert_sign_text)
+
+        self.layout.addWidget(self.run_all_button, 0, 0, 3, 4)
+        self.layout.addWidget(self.save_preproc_check, 3, 0, 1, 2)
+        self.layout.addWidget(self.clear_cache_check, 3, 2, 1, 2)
+        self.layout.addWidget(self.do_CAR_check, 4, 0, 1, 2)
+        self.layout.addWidget(self.invert_sign_check, 4, 2, 1, 2)
 
         self.setLayout(self.layout)
 

kilosort/gui/sorter.py

@@ -52,17 +52,13 @@ def run(self):
             try:
                 logger.info(f"Kilosort version {kilosort.__version__}")
                 logger.info(f"Sorting {self.data_path}")
+                clear_cache = settings['clear_cache']
+                if clear_cache:
+                    logger.info('clear_cache=True')
                 logger.info('-'*40)
 
                 tic0 = time.time()
 
-                # TODO: make these options in GUI
-                do_CAR=True
-                invert_sign=False
-
-                if not do_CAR:
-                    logger.info("Skipping common average reference.")
-
                 if probe['chanMap'].max() >= settings['n_chan_bin']:
                     raise ValueError(
                         f'Largest value of chanMap exceeds channel count of data, '
@@ -74,9 +70,13 @@ def run(self):
                 data_dtype = settings['data_dtype']
                 device = self.device
                 save_preprocessed_copy = settings['save_preprocessed_copy']
+                do_CAR = settings['do_CAR']
+                invert_sign = settings['invert_sign']
+                if not do_CAR:
+                    logger.info("Skipping common average reference.")
 
                 ops = initialize_ops(settings, probe, data_dtype, do_CAR,
-                                    invert_sign, device, save_preprocessed_copy)
+                                     invert_sign, device, save_preprocessed_copy)
                 # Remove some stuff that doesn't need to be printed twice,
                 # then pretty-print format for log file.
                 ops_copy = ops.copy()
@@ -94,7 +94,7 @@ def run(self):
                 torch.random.manual_seed(1)
                 ops, bfile, st0 = compute_drift_correction(
                     ops, self.device, tic0=tic0, progress_bar=self.progress_bar,
-                    file_object=self.file_object
+                    file_object=self.file_object, clear_cache=clear_cache
                     )
 
                 # Check scale of data for log file
@@ -113,7 +113,7 @@ def run(self):
                 # Sort spikes and save results
                 st, tF, Wall0, clu0 = detect_spikes(
                     ops, self.device, bfile, tic0=tic0,
-                    progress_bar=self.progress_bar
+                    progress_bar=self.progress_bar, clear_cache=clear_cache
                     )
 
                 self.Wall0 = Wall0
@@ -123,7 +123,7 @@ def run(self):
 
                 clu, Wall = cluster_spikes(
                     st, tF, ops, self.device, bfile, tic0=tic0,
-                    progress_bar=self.progress_bar
+                    progress_bar=self.progress_bar, clear_cache=clear_cache
                     )
                 ops, similar_templates, is_ref, est_contam_rate, kept_spikes = \
                     save_sorting(ops, results_dir, st, clu, tF, Wall, bfile.imin, tic0)