def update_graph(tiff_mode, filename, graph_idx, curr_neuron):
background = templates_tiff[graph_idx] if tiff_mode else templates[
graph_idx]
if curr_neuron == -1:
new_graph = int_func.format_fig(background,
title=filename.partition(".")[0],
zoom=False,
is_tiff_mode=tiff_mode)
else:
if not tiff_mode:
background = np.where(
np.logical_and(footprints[graph_idx][curr_neuron] >= 0.01,
background != 255),
# to keep com dot
cols[0],
background)
new_graph = int_func.format_fig(
background,
title=filename.partition(".")[0],
zoom=True,
zoom_ratio=zoom_ratio,
center_coords_x=int(coms[graph_idx][curr_neuron][1]),
center_coords_y=int(coms[graph_idx][curr_neuron][0]),
is_tiff_mode=tiff_mode)
return new_graph
def update_graphs(fn0, fn1, fn2, fn3, fn4, fn5, fn6, fn7, fn8, fn9, fn10, fn11,
cd0, cd1, cd2, cd3, cd4, cd5, cd6, cd7, cd8, cd9, cd10, cd11,
click_save, suggestion_mode, tiff_mode, no_match_update,
st0, st1, st2, st3, st4, st5, st6, st7, st8, st9, st10, st11,
state_cells_left):
fnames = [fn0, fn1, fn2, fn3, fn4, fn5, fn6, fn7, fn8, fn9, fn10, fn11]
cdata = [cd0, cd1, cd2, cd3, cd4, cd5, cd6, cd7, cd8, cd9, cd10, cd11]
states = [st0, st1, st2, st3, st4, st5, st6, st7, st8, st9, st10, st11]
updates = [no_update] * num_plots_max
new_sugg = False
cells_left_message = state_cells_left
ctx = dash.callback_context
reset = no_update
global cell_matching_df
global templates
global orig_templates
if ctx.triggered:
global current_selected
trigger_action = ctx.triggered[0]['prop_id'].split('.')[1]
trigger_firstarg = ctx.triggered[0]['prop_id'].split('.')[0]
# 1. SAVE BUTTON CLICKED
if trigger_action == "n_clicks" and trigger_firstarg == "save_button":
print("***callback: SAVE")
if not filenames[0]: # no files uploaded and/or nothing highlighted
cells_left_message = "upload files first"
else:
# get index, update cells of matched neurons and recompute best options for other neurons
col = cell_matching_df[fnames[0].partition(".")[0]]
row_idx = cell_matching_df[col == current_selected[0]].index.values[0]
cell_matching_df.at[row_idx, "confirmed"] = 1
for graph_idx, filename in enumerate(filenames):
if filename:
cell_matching_df.at[row_idx, filename] = current_selected[graph_idx]
cell_matching_df.to_csv(UPLOAD_DIRECTORY + "\\" + filename_result_csv, index=False, header=True, sep=';')
print("csv saved @ ", UPLOAD_DIRECTORY + "\\" + filename_result_csv)
# output number of cells left to match
num_cells_left = sum(cell_matching_df["confirmed"] == 0)
cells_left_message = str(num_cells_left) + " cells left to confirm"
# color neuron in graphs
color = cell_matching_df.at[row_idx, "color"]
for graph_idx, filename in enumerate(filenames):
print(graph_idx)
curr_neuron = current_selected[graph_idx]
if not filename or curr_neuron == -1:
continue
print(curr_neuron, color)
templates[graph_idx] = np.where(
np.logical_and(footprints[graph_idx][curr_neuron] >= 0.01, templates[graph_idx] != 255),
cols[color], templates[graph_idx])
updates[graph_idx] = update_graph(tiff_mode, filename, graph_idx, -1)
print(cells_left_message)
new_sugg = suggestion_mode
reset = True
# SUGGEST RANDOM NEURON MATCHING
if (ctx.triggered[0]['prop_id'] == "suggestion_mode_button.active") or new_sugg:
if ctx.triggered[0]['value'] or new_sugg:
print("new neuron matching suggested")
if isinstance(cell_matching_df, pd.DataFrame):
col = cell_matching_df["confirmed"]
row_idx = cell_matching_df[col == 0].index.values[0]
for graph_idx, filename in enumerate(fnames):
if not filename:
continue
print(filename, fnames)
curr_neuron = cell_matching_df.iloc[row_idx][filename.partition(".")[0]]
if graph_idx > 0:
clf_input = cell_matching_df.iloc[row_idx][filename.partition(".")[0] + 'feature_vals']
if not isinstance(clf_input, list):
clf_input = json.loads(clf_input)
print('clf_input ', clf_input)
curr_neuron = get_predicted_neuron(clf, clf_input[:3])
if isinstance(curr_neuron, list):
print("list of tuples ", curr_neuron)
curr_neuron = curr_neuron[0][1]
print(curr_neuron);
current_selected[graph_idx] = curr_neuron
updates[graph_idx] = update_graph(tiff_mode, filename, graph_idx, curr_neuron)
print(int(coms[graph_idx][curr_neuron][0]), int(coms[graph_idx][curr_neuron][1]))
# IMAGE MODE CHANGED
elif ctx.triggered[0]['prop_id'] == "tiff_mode_button.active":
print("***callback: IMAGE MODE tiff CHANGED to: ", tiff_mode)
for graph_idx, filename in enumerate(fnames):
if not filename:
continue
print(filename, fnames, current_selected[graph_idx])
updates[graph_idx] = update_graph(tiff_mode, filename, graph_idx, current_selected[graph_idx])
# NO MATCH BUTTON UPDATED
elif ctx.triggered[0]['prop_id'] == "no_match_update_graph.children":
print("***callback: GRAPH UPDATED because of NO MATCH BUTTON click: ")
for graph_idx, filename in enumerate(fnames):
if not filename or not graph_idx: # not ref
continue
current_selected[graph_idx] = -1 if no_match[graph_idx - 1] else current_selected[graph_idx]
updates[graph_idx] = update_graph(tiff_mode, filename, graph_idx, current_selected[graph_idx])
# 2 GRAPH CLICKED
elif trigger_action == "clickData":
trigger_graph_id = int(trigger_firstarg.rsplit('_', 1)[-1])
print("***callback: IMAGE CLICKED trigger_id: ", trigger_graph_id)
click_data = cdata[trigger_graph_id]
print(click_data)
y = click_data['points'][0]['x']
x = click_data['points'][0]['y']
print("click on something", x, y, fnames[trigger_graph_id], np.amax(pixel_ownerships[trigger_graph_id]))
neuron_id = pixel_ownerships[trigger_graph_id][x, y]
print("clicked on neuron: ", neuron_id, "trigger graph: ", trigger_graph_id)
if neuron_id != -1 and neuron_id != current_selected[trigger_graph_id]:
current_selected[trigger_graph_id] = neuron_id
# reselection of matching -> update highlighted neuron (of a non-reference graph)
if trigger_graph_id != 0:
if not no_match[trigger_graph_id-1]: # cannot update graph if 'no match' button is pressed!
print("newly selected: ", current_selected[trigger_graph_id])
updates[trigger_graph_id] = update_graph(tiff_mode, fnames[trigger_graph_id], trigger_graph_id,
current_selected[trigger_graph_id])
# new ref neuron clicked -> update highlighted neuron in ref and non-ref graphs
if trigger_graph_id == 0 and not suggestion_mode:
try:
print(list(cell_matching_df.columns), fnames[trigger_graph_id].partition(".")[0])
except AttributeError:
print("Compute or upload matchings before clicking neurons!")
col = cell_matching_df[fnames[trigger_graph_id].partition(".")[0]]
row_idx = cell_matching_df[col == neuron_id].index.values[0]
print("row index is: ", row_idx)
for graph_idx, filename in enumerate(fnames):
if not filename:
continue
print(filename, fnames)
curr_neuron = cell_matching_df.iloc[row_idx][filename.partition(".")[0]]
if graph_idx > 0:
clf_input = cell_matching_df.iloc[row_idx][filename.partition(".")[0] + 'feature_vals']
if not isinstance(clf_input, list):
clf_input = json.loads(clf_input)
print('clf_input ', clf_input)
curr_neuron = get_predicted_neuron(clf, clf_input[:3])
print(curr_neuron)
current_selected[graph_idx] = curr_neuron
updates[graph_idx] = update_graph(tiff_mode, filename, graph_idx, curr_neuron)
print(int(coms[graph_idx][curr_neuron][0]), int(coms[graph_idx][curr_neuron][1]))
print("click on nothing?")
# 3 UPLOAD BUTTON
elif trigger_action == 'filename':
print("***callback: LOAD FILE")
print(ctx.triggered)
trigger_graph_id = int(trigger_firstarg.rsplit('_', 1)[-1])
print("loading ", fnames[trigger_graph_id])
updates[trigger_graph_id] = int_func.format_fig(
getdata_from_npyfile(fnames[trigger_graph_id], trigger_graph_id)[1],
title=fnames[trigger_graph_id].partition(".")[0])
print("done loading file")
for i in range(num_plots_max):
if i != trigger_graph_id:
updates[i] = no_update
else:
pass
print(cells_left_message)
return tuple(updates) + (cells_left_message,) + (reset,)
# initialization of global variables
num_plots_max = 12
cell_matching_df = []
filenames = [''] * num_plots_max
sessions_data = [None] * num_plots_max
footprints = [None] * num_plots_max
orig_templates = [None] * num_plots_max
templates = [None] * num_plots_max
templates_tiff = [None] * num_plots_max
coms = [None] * num_plots_max
current_selected = [-1] * num_plots_max
no_match = [False] * (num_plots_max - 1)
pixel_ownerships = [np.full((512, 512), -1)] * num_plots_max
empty_template = [[0] * 512] * 512
figs = [int_func.format_fig(empty_template)] * num_plots_max
clf = pickle.load(open(classifier_path, 'rb'))
# **********************************************************************************************************************
# DASHBOARD
# **********************************************************************************************************************
print("the app reset, all data must be reloaded!")
app = dash.Dash(external_stylesheets=[dbc.themes.BOOTSTRAP])
# **********************************************************************************************************************
# LAYOUT (see content_layout.py for more details)
# **********************************************************************************************************************
app.layout = dbc.Container(
# initialization of global variables
cell_matching_df = []
filenames = [''] * 6
sessions_data = [None] * 6
footprints = [None] * 6
orig_templates = [None] * 6
templates = [None] * 6
templates_tiff = [None] * 6
coms = [None] * 6
current_selected = [-1] * 6
no_match = [False] * 5
pixel_ownerships = [np.full((512, 512), -1)] * 6
empty_template = [[0] * 512] * 512
figs = [int_func.format_fig(empty_template)] * 6
clf = pickle.load(open(classifier_path, 'rb'))
# **********************************************************************************************************************
# DASHBOARD
# **********************************************************************************************************************
print("the app reset, all data must be reloaded!")
app = dash.Dash(external_stylesheets=[dbc.themes.BOOTSTRAP])
# **********************************************************************************************************************
# LAYOUT (see content_layout.py for more details)
# **********************************************************************************************************************
app.layout = dbc.Container([
dbc.Row(dbc.Col(html.Div("CELL MATCHING ACROSS SESSIONS"), width=12),