Fix for the probabilityLeft of some of the bugged BiasedCW sessions

src/iblphotometry/BCW_bug_solution_KB.py

@@ -0,0 +1,136 @@
+""" 
+KB 2025-05-25
+Solution proposed on Oct2024
+
+Code to fix the BCW bug in the IBL photometry sessions
+
+This code is used to fix the bug in the labelling of some of the photometry BCW sessions - the task ran correctly, but the 
+labelling was incorrect for the probabilityLeft: 
+ - BCW block 1 - 50/50 = correct
+ - BCW block 2 - 20/80 or 80/20 = correct
+ - BCW block 3 - 80/20 or 20/80 = incorrect - it starts shifting 0.2 and 0.8 for each consecutive trial
+
+""" 
+
+#%%
+import numpy as np
+import pandas as pd
+import matplotlib.pyplot as plt
+from one.api import ONE #always after the imports 
+one = ONE(cache_rest=None, mode='remote') 
+
+#functions 
+def get_eid(): 
+    # eids = one.search(project='ibl_fibrephotometry') 
+    # use example eid with bug
+    eid = 'd3ac8f32-cfba-4f48-b513-9d998ab0ae72'
+    ref = one.eid2ref(eid)
+    print(eid)
+    print(ref) 
+    try:
+        # Try to load the trials directly
+        a = one.load_object(eid, 'trials')
+        trials = a.to_df()
+        trials['trialNumber'] = range(1, len(trials) + 1) 
+        trials["mouse"] = ref.subject
+        trials["date"] = ref.date
+        trials["eid"] = eid 
+        df_trials = trials
+    except: 
+        print("session not found")
+
+    return df_trials, eid
+
+df_trials, eid = get_eid()
+
+plt.plot(df_trials.probabilityLeft, alpha=0.5)
+
+
+#%%
+""" LOAD TRIALS """
+def load_trials_updated(eid=eid): 
+    trials = one.load_object(eid, 'trials')
+    ref = one.eid2ref(eid)
+    subject = ref.subject
+    session_date = str(ref.date) 
+    if len(trials['intervals'].shape) == 2: 
+        trials['intervals_0'] = trials['intervals'][:, 0]
+        trials['intervals_1'] = trials['intervals'][:, 1]
+        del trials['intervals']  # Remove original nested array 
+    df_trials = pd.DataFrame(trials) 
+    idx = 2
+    new_col = df_trials['contrastLeft'].fillna(df_trials['contrastRight']) 
+    df_trials.insert(loc=idx, column='allContrasts', value=new_col) 
+    # create allSContrasts 
+    df_trials['allSContrasts'] = df_trials['allContrasts']
+    df_trials.loc[df_trials['contrastRight'].isna(), 'allSContrasts'] = df_trials['allContrasts'] * -1
+    df_trials.insert(loc=3, column='allSContrasts', value=df_trials.pop('allSContrasts'))
+    df_trials[["subject", "date", "eid"]] = [subject, session_date, eid]    
+    df_trials["reactionTime"] = df_trials["firstMovement_times"] - df_trials["stimOnTrigger_times"]
+    df_trials["responseTime"] = df_trials["response_times"] - df_trials["stimOnTrigger_times"] 
+    df_trials["quiescenceTime"] = df_trials["stimOnTrigger_times"] - df_trials["intervals_0"] 
+    df_trials["trialTime"] = df_trials["intervals_1"] - df_trials["intervals_0"]  
+
+    try: 
+        dataset_task_settings = one.load_dataset(eid, '_iblrig_taskSettings.raw.json')  
+        values = dataset_task_settings.get('LEN_BLOCKS', 'Key not found') 
+        # values gives the block length 
+        # example for     eid = 'd3ac8f32-cfba-4f48-b513-9d998ab0ae72'
+        # values = [90, 41, 65, 62, 30, 31, 64, 82, 33, 80, 70, 62, 40, 70, 22, 72, 60, 22, 30, 53, 51, 28, 31, 28, 22, 41, 72, 72, 51, 70, 24, 30, 55, 78, 39, 53, 23, 53, 25, 21, 48, 97]
+
+
+        values_sum = np.cumsum(values) 
+
+        # Initialize a new column 'probL' with NaN values
+        df_trials['probL'] = np.nan
+
+        # Set the first block (first `values_sum[0]` rows) to 0.5
+        df_trials.loc[:values_sum[0]-1, 'probL'] = 0.5 
+
+
+        df_trials.loc[values_sum[0]:values_sum[1]-1, 'probL'] = df_trials.loc[values_sum[0], 'probabilityLeft']
+
+        previous_value = df_trials.loc[values_sum[1]-1, 'probabilityLeft'] 
+
+
+        # Iterate over the blocks starting from values_sum[1]
+        for i in range(1, len(values_sum)-1):
+            print("i = ", i)
+            start_idx = values_sum[i]
+            end_idx = values_sum[i+1]-1
+            print("start and end _idx = ", start_idx, end_idx)
+
+            # Assign the block value based on the previous one
+            if previous_value == 0.2:
+                current_value = 0.8
+            else:
+                current_value = 0.2
+            print("current value = ", current_value)
+
+
+            # Set the 'probL' values for the current block
+            df_trials.loc[start_idx:end_idx, 'probL'] = current_value
+
+            # Update the previous_value for the next block
+            previous_value = current_value
+
+        # Handle any remaining rows after the last value_sum block
+        if len(df_trials) > values_sum[-1]:
+            df_trials.loc[values_sum[-1] + 1:, 'probL'] = previous_value
+
+        # plt.plot(df_trials.probabilityLeft, alpha=0.5)
+        # plt.plot(df_trials.probL, alpha=0.5)
+        # plt.title(f'behavior_{subject}_{session_date}_{eid}')
+        # plt.show() 
+    except: 
+        pass 
+
+    df_trials["trialNumber"] = range(1, len(df_trials) + 1) 
+    return df_trials, subject, session_date
+
+df_trials, subject, session_date = load_trials_updated(eid) 
+
+plt.plot(df_trials.probabilityLeft, alpha=0.5) #bug
+plt.plot(df_trials.probL, alpha=0.85, color='red') #bug solved
+
+# %%

src/iblphotometry/BCW_bugged_eids_list.csv

@@ -0,0 +1,83 @@
+eid
+488cb758-888a-4664-bc8e-2e0624507ce4
+61073b6e-842e-440e-bf59-b3197065bfa1
+24415d91-8a3a-440b-beee-2d4388a4b908
+f416af34-121d-419c-8d27-4e95f8648012
+b46d38e4-cba5-4d2f-b975-93542a452317
+9c86d1db-cb3f-4bfe-bdd4-450c25b1b779
+385d2bc7-02e6-4d65-b20a-68ea9bd85e7c
+56044f43-8550-4b36-ae5d-0a32e627c0fd
+65b43acb-afa5-4590-a400-f7f4c79788ab
+cd9d071e-c798-4900-891f-b65640ec22b1
+25ff1634-88e8-47bf-9f3a-f527d0f94444
+5104ba1a-9c35-470a-8f6d-69a07349e19b
+d3ac8f32-cfba-4f48-b513-9d998ab0ae72
+4f9da1a0-5465-410b-83f4-e852cfe61107
+40a2c7c5-bdce-4aae-ab2f-e088d82e58d3
+8996a0d0-3b23-4d7e-92a4-8272b4bb5a9f
+be3208c9-43de-44dc-bdc6-ff8963464f98
+5b7c1f53-0c33-4ac3-8838-2926f46abf13
+490f4be9-de15-4251-aae4-2e059bbc03c9
+cb7994bf-f063-46ea-bd05-aa6e9a2fc7ef
+274c4683-ef2a-471a-841f-af634446ec8d
+e69221c7-b533-42dc-8631-000279a45a70
+55d7f234-cc47-44a6-bb36-086c854eff00
+57c0be0d-5010-4257-bab5-dbf7eca081f3
+4b42f52f-3e60-41f6-869a-465ee241d0e8
+86e88a0d-a637-4bf4-a215-ef6013277234
+cb21401e-dfe6-4e81-b6d6-df14107a45c3
+5e847644-4e5b-4074-bb48-8218c0fa76fd
+cc0b6954-3030-488e-8d10-a1ba421d655a
+9fac6cf0-8dc6-456d-9dd5-48be737df408
+85764273-c611-491f-b4f5-de127dde1cf1
+3df9f2f6-708a-48aa-b033-050750e62102
+c9961dce-e106-49b7-8892-9d7dc22675be
+4590ea45-e39a-45cd-925c-542d4a5eb97b
+5ca99111-68bd-40cf-a582-7956a9057b0d
+708a3e0e-d8d7-43a8-b182-6a4192e4dc80
+7757b591-7188-4418-8a9c-ae6cf17cb82c
+a54d305f-2089-47ac-8e75-134db4fd3ed9
+6aefe272-b19a-41df-9adc-5af6cedb8dab
+e7c37ff1-e3ee-4e82-813f-f408af91730c
+d226644e-d10b-4142-9e5e-17aa91f378d6
+8404a92c-65d1-4923-9efd-13155ec1c8db
+48e60716-5850-4262-b9cd-24ffb6a7189f
+09d25ccb-b0b5-4e89-81ff-974128934390
+82114940-6291-4e2c-8d3e-4f3b73c081de
+272cfbd8-d1b7-4a3b-aa13-dfc514667b46
+87809791-48fe-47fd-9973-03ddb860f8cd
+b5e94727-4842-4e81-830a-2b48b60ecf0b
+46fe69ff-d001-4608-a15e-d5e029c14fc3
+31ca8cb2-92a7-4608-a998-104ec5db1c54
+89b8ef70-e620-49c2-a0f7-09890ba9fc0e
+beae9e93-2826-4205-b9f9-376dd785dad0
+9f19fbe9-766b-4324-b2bf-a55c7131949b
+4be05d77-d7fd-40f4-a60a-0b9269f4bcb1
+628d24c4-6305-4125-b5bd-4723a03a5be8
+fc8963f9-7282-4a05-a4a6-41ce0a1541df
+8062e7d8-f770-47bc-94f5-d02691d8b02a
+820ece0e-aa62-44fe-a881-dbd51e49748e
+0f5f08e1-b2b6-4306-b4de-c89b1156581d
+c4dcc2bd-e701-4a28-bfaa-d796320201bd
+0e951224-308e-4f91-96c7-58414d335f10
+60246d4e-2386-443a-a5d6-94ac015aaf06
+382c9380-a2f7-411f-949d-a6df6c00747f
+f0eb3ff8-873e-4bce-8082-7c8e28b47edb
+950565f4-1054-4a58-b5c9-8fa47ff42fe6
+0b842869-94c4-4248-a302-a31859136ad2
+65aa40c1-3400-4717-8854-4cfba85b5899
+e316594f-bcdb-42ba-b8e8-68d80a456a34
+67231166-8980-4146-ad6d-3d7bee327ccf
+76f443fb-a06d-4ffd-9725-23d5c0d40386
+9e8c2697-f5f7-42f6-b483-5b9af9c1652e
+537a8269-3cdb-4738-9b0f-1134cd4e3d0d
+160cff05-de6b-4070-9eaa-0bbbcb0a71d5
+66d51ecb-b6a0-4293-84be-d4ccfb84ada8
+8cfa1e92-3e7f-409f-8347-01a36e98b27f
+f36ae1d4-20e0-47ee-b4bd-9649810d87e0
+a2e65766-0f40-46dd-91cf-b56540cb033b
+4783a2d1-2761-4c66-93fe-71b310b49261
+a4df6acb-bf2c-4362-80c1-a1a207aaf86e
+69cb5d57-109a-4455-b469-9cc235812cfb
+df899930-8172-400a-a7f0-07061382a81b
+714faf78-b4a5-4aff-9504-6f270dd139be