recipe(meshes) shift subject meshes on embedding position

dico_toolbox/mesh.py

@@ -124,4 +124,4 @@ def apply_Talairach_to_mesh(mesh, dxyz, rotation, translation, flip=False):
     if flip:
         flip_mesh(mesh)
 
-    return mesh
+    return mesh

dico_toolbox/recipes/__init__.py

@@ -1 +1 @@
-from .meshes import mesh_of_average, mesh_of_averages, mesh_one_point_cloud, mesh_of_point_clouds
+from .meshes import mesh_of_average, mesh_of_averages, mesh_one_point_cloud, mesh_of_point_clouds, shift_meshes_in_embedding

dico_toolbox/recipes/meshes.py

@@ -71,7 +71,6 @@ def mesh_one_point_cloud(data):
     tal = data["talairach"]
     flip = data['flip']
     align = data["align"]
-    shift = data["shift"]
     meshing_parameters = data['meshing_parameters']
 
     # generate mesh
@@ -110,32 +109,26 @@ def _parse_pool_results(results):
     return meshes
 
 
-def mesh_of_point_clouds(pcs, pre_transformation=None, flip=False, post_transformation=None, embedding=None, embedding_scale=1, **meshing_parameters):
+def mesh_of_point_clouds(pcs, pre_transformation=None, flip=False, post_transformation=None, **meshing_parameters):
     """Build the mesh of the pointclouds.
 
     Args:
         pcs (dict): the point clouds
         pre_transformation (collection of dict, optional): This transformation is applied before flip. keys = {dxy, rot, tra}. Defaults to None.
         flip (bool, optional): flip the data. Defaults to False.
         post_transformation (collection of dict, optional): This transformation is applied after flip. keys = {rot, tra}. Defaults to None.
-        embedding (DataFrame, optional): The embedding. Defaults to None.
-        embedding_scale (int, optional): Scale for the final offset of the mesh. Defaults to 1.
 
     Returns:
         _type_: _description_
     """
     data = []
     for name, pc in pcs.items():
-        shift = None
-        if embedding is not None:
-            shift = embedding.loc[name].values*embedding_scale
         data.append(dict(
             name=name,
             pc=pc,
             talairach=pre_transformation,  # {dxy, rot, tra}
             flip=flip,
             align=post_transformation,  # {rot, tra},
-            shift=shift,  # [x,y,z]
             meshing_parameters=meshing_parameters
         ))
 
@@ -147,3 +140,32 @@ def mesh_of_point_clouds(pcs, pre_transformation=None, flip=False, post_transfor
         res = _parse_pool_results(res)
 
     return res
+
+
+def shift_meshes_in_embedding(meshes:dict, embedding, scale=1):
+    """Shift the meshes to an amount proportional to their position in the embedding.
+    
+    Args:
+        -meshes:dict(name,ndarray) name:point-cloud dictionary
+        -embedding:pandas.DataFrame the embedding of the point-clouds (e.g. isomap axis)
+        -scale: multiplicative factor for the actual shift.
+    
+    Return:
+        -name:dict(shifted_meshes)
+    """
+    shifted_meshes = dict()
+
+    for name, mesh in tqdm(meshes.items(), desc="shifting"):
+        shift_v = np.zeros(3)
+        assert len(embedding.shape)<3, "For displaying, the embedding dimension must be < 3"
+
+        if len(embedding.shape) == 1:
+            shift_v[0] = embedding.loc[name]
+        else:
+            for i,v in enumerate(embedding.loc[name]):
+                shift_v[i]=v
+
+
+        shifted_mesh = shift_aims_mesh(mesh, shift_v, scale=scale)
+        shifted_meshes[name] = shifted_mesh
+    return shifted_meshes