Reverting back to storage of raw data in a separate store

src/funki/app.py

@@ -29,6 +29,7 @@
 app.layout = html.Div(
     children=[
         dcc.Store(id='data', storage_type=storage_type),
+        dcc.Store(id='raw', storage_type=storage_type),
         html.Div(
             html.Img(
                 src='assets/logos/funki_logo.svg',

src/funki/pages/data.py

@@ -213,6 +213,7 @@
 
 @callback(
     Output('data', 'data', allow_duplicate=True),
+    Output('raw', 'data', allow_duplicate=True),
     Input('upload-data', 'contents'),
     State('upload-data', 'filename'),
     State('separator-data', 'value'),
@@ -225,19 +226,20 @@ def load_data(content, filename, sep):
     df = parse_contents(content, filename, sep=sep).astype(np.float32)
 
     serial = dataframe_to_serial(df)
-    serial.update({'raw': dataframe_to_serial(df.copy())})
 
-    return serial
+    return serial, serial.copy()
 
 @callback(
     Output('data', 'data', allow_duplicate=True),
+    Output('raw', 'data', allow_duplicate=True),
     Input('upload-obs', 'contents'),
     State('upload-obs', 'filename'),
     State('data', 'data'),
+    State('raw', 'data'),
     State('separator-obs', 'value'),
     prevent_initial_call=True
 )
-def load_obs(content, filename, data, sep):
+def load_obs(content, filename, data, raw, sep):
     if filename is None:
         raise PreventUpdate
 
@@ -246,8 +248,9 @@ def load_obs(content, filename, data, sep):
 
     serial = dataframe_to_serial(parse_contents(content, filename, sep=sep))
     data.update({'obs': serial})
+    raw.update({'obs': serial.copy()})
 
-    return data
+    return data, raw
 
 @callback(
     Output('table-data', 'columns'),

src/funki/pages/difexp.py

@@ -158,20 +158,22 @@ def update_group_selector(data, obs_var, va, vb):
     Output('plot-dex', 'figure'),
     Input('apply-dex', 'n_clicks'),
     State('data', 'data'),
+    State('raw', 'data'),
     State('obs-selector', 'value'),
     State('group-selector-a', 'value'),
     State('group-selector-b', 'value'),
     prevent_initial_call=True
 )
-def apply_dex(n_clicks, data, obs_var, groups_a, groups_b):
+def apply_dex(n_clicks, data, raw, obs_var, groups_a, groups_b):
     if data is None:
         raise PreventUpdate
 
     if not all([obs_var, groups_a, groups_b]):
         raise PreventUpdate
 
-    dset_raw = serial_to_dataset(data['raw'])
-    dset_raw.obs = serial_to_dataframe(data['obs'])
+    dset_raw = serial_to_dataset(raw)
+    if 'obs' in raw.keys():
+        dset_raw.obs = serial_to_dataframe(raw['obs'])
 
     # Retrieve var as dataframe if available
     if 'var' in data.keys():
@@ -182,7 +184,7 @@ def apply_dex(n_clicks, data, obs_var, groups_a, groups_b):
 
     # Re-applying filters to raw data if any
     if 'uns' in data:
-        uns = uns = data['uns']['funki']['sc_trans_filter']
+        uns = data['uns']['funki']['sc_trans_filter']
         dset_raw = fpp.sc_trans_filter(
             dset_raw,
             min_genes=uns['min_genes'] if 'min_genes' in uns.keys() else None,

src/funki/pages/norm.py

@@ -180,6 +180,7 @@ def apply_filter(n_clicks, data, max_genes, min_genes, mito_pct):
         raise PreventUpdate
 
     dset = serial_to_dataset(data)
+
     dset = fpp.sc_trans_filter(
         dset,
         min_genes=min_genes,

src/funki/utils/utils.py

@@ -68,10 +68,9 @@ def serial_to_dataset(data):
         if k in data.keys()
     })
 
-    if 'raw' in data.keys():
-        kwargs.update({'raw': serial_to_dataset(data['raw'])})
+    dset = DataSet(df.astype(np.float32), **kwargs)
 
-    return DataSet(df.astype(np.float32), **kwargs)
+    return dset
 
 def dataset_to_serial(dset):
     data = dataframe_to_serial(dset.to_df())
@@ -82,10 +81,6 @@ def dataset_to_serial(dset):
         if not getattr(dset, k).empty
     })
 
-    if 'raw' in dset.__dict__.keys():
-        df = pd.DataFame(dset.raw.X, columns=dset.raw.var)
-        data['raw'] = dataframe_to_serial(df)
-
     attrs = {}
 
     for k in ('obsm', 'varm', 'obsp', 'varp'):