davy's bug fixes

Author: mansouralawi

examples/wrappers/3_alignment_evaluation.py

@@ -11,7 +11,7 @@
 args_file_path = '/path/to/your/parameter/file.json'
 
 # Load the configuration settings from the specified file.
-args.load_config(args_file_path)
+args.load_params(args_file_path)
 
 # Step 2: Additional Configuration for Alignment Evaluation
 # ================================================
@@ -47,7 +47,10 @@
 # Extract the coordinates.
 for fov in fovs_to_analyze:
     for code in codes_to_analyze:
-        measure_round_alignment_NCC(args=args,code=code, fov=fov)
+        try:
+            measure_round_alignment_NCC(args=args,code=code, fov=fov)
+        except Exception as e:
+            print(f"Error measuring alignment for Code {code}, FOV {fov}: {e}")
 
 
 # Step 4: Alignment Evaluation and Confidence Interval Calculation

exm/align/align_eval.py

@@ -115,7 +115,7 @@ def plot_alignment_evaluation(args: Args, fov: int, percentile: int = 95, save_f
     """
 
     def load_eval_data(args: Args, fov: int) -> dict:
-        eval_file = os.path.join(args.processed_data_path, "alignment_evaluation", f"FOV{fov}", f"NCC_fov{fov}.json")
+        eval_file = os.path.join(args.processed_data_path, "alignment_evaluation", f"FOV{fov}", f"NCC_FOV{fov}.json")
         with open(eval_file, 'r') as f:
             eval_data = json.load(f)
         return eval_data
@@ -189,7 +189,7 @@ def bootstrap_mean_ci(data: np.array, n_bootstrap_samples: int, ci: float) -> (f
         return lower_bound, upper_bound
 
     ci_data = {}
-    eval_file_path = os.path.join(args.processed_data_path, "alignment_evaluation", f"ROI{roi}", f"NCC_ROI{roi}.json")
+    eval_file_path = os.path.join(args.processed_data_path, "alignment_evaluation", f"FOV{fov}", f"NCC_FOV{fov}.json")
 
     try:
         with open(eval_file_path, 'r') as f:
@@ -203,7 +203,7 @@ def bootstrap_mean_ci(data: np.array, n_bootstrap_samples: int, ci: float) -> (f
             percentile_value = np.percentile(eval_array, percentile_filter)
             eval_filtered = eval_array[eval_array <= percentile_value]
 
-            lower_bound, upper_bound = bootstrap_mean_ci(eval_filtered, len(eval_data)//10, ci)
+            lower_bound, upper_bound = bootstrap_mean_ci(eval_filtered, len(eval_filtered)//10, ci)
             ci_data[key] = {
                 "lower_bound": lower_bound,
                 "upper_bound": upper_bound

exm/puncta/consolidate.py

@@ -204,6 +204,10 @@ def find_matching_points(point_cloud1, point_cloud2, distance_threshold=8):
                         point_cloud1 = np.asarray([x["position"] for x in reference])
                         point_cloud2 = np.asarray([x["position"] for x in new])
 
+                        if len(point_cloud2) == 0 or len(point_cloud1) == 0:
+                            logger.warning(f"Code {code} for FOV {fov} has no puncta.")
+                            continue
+
                         pairs = find_matching_points(point_cloud1, point_cloud2)
 
                         for pair in pairs:

exm/segmentation/segmentation.py

@@ -8,7 +8,7 @@
 from cellpose import models,utils,plot,io
 
 
-def segment_3d(volume,model,downsample=False,chan=0,chan2=0,flow_threshold=0.4,cellprob_threshold=0,do_3D=True):
+def segment_3d(volume,model,downsample=False,chan=0,chan2=0,diameter=30,flow_threshold=0.4,cellprob_threshold=0,do_3D=True):
     r"""
     Performs 3D segmentation on the given volume using the provided model.
 
@@ -22,6 +22,8 @@ def segment_3d(volume,model,downsample=False,chan=0,chan2=0,flow_threshold=0.4,c
     :type chan: int
     :param chan2: The second channel to use for segmentation. Default is 0.
     :type chan2: int
+    :param diameter: Estimated diameter of objects to segment. Set to 0 to auto-estimate. Default is 30.
+    :type diameter: float
     :param flow_threshold: The flow threshold for the segmentation. Default is 0.4.
     :type flow_threshold: float
     :param cellprob_threshold: The cell probability threshold for the segmentation. Default is 0.
@@ -36,9 +38,10 @@ def segment_3d(volume,model,downsample=False,chan=0,chan2=0,flow_threshold=0.4,c
         volume = ndimage.zoom(volume, (1,1,0.25,0.25), order= 1)    
     masks, flows, styles = model.eval(volume, 
                                   channels=[chan, chan2],
-                                  diameter=model.diam_labels,
+                                  diameter=diameter,
                                   flow_threshold=flow_threshold,
                                   cellprob_threshold=cellprob_threshold,
+                                  z_axis=0,
                                   do_3D=do_3D
                                   )
     if downsample: