hanhou/patchseq
0
1"""2Panel-based visualization tool for navigating and visualizing patch-seq NWB files.3 4To start the app, run:5panel serve panel_nwb_viz.py --dev --allow-websocket-origin=codeocean.allenneuraldynamics.org --title "Patch-seq Data Explorer" # noqa: E5016"""7 8import logging9import re10 11import numpy as np12import pandas as pd13import panel as pn14import param15from bokeh.io import curdoc16from bokeh.layouts import column as bokeh_column17from bokeh.models import (18 BoxZoomTool,19)20from bokeh.plotting import figure21 22from LCNE_patchseq_analysis.data_util.metadata import load_ephys_metadata23from LCNE_patchseq_analysis.data_util.nwb import PatchSeqNWB24from LCNE_patchseq_analysis.pipeline_util.s3 import (25 S3_PUBLIC_URL_BASE,26 get_public_url_cell_summary,27 get_public_url_sweep,28 load_efel_features_from_roi,29)30from LCNE_patchseq_analysis.population_analysis.spikes import (31 extract_representative_spikes,32)33 34from components.scatter_plot import ScatterPlot35from components.spike_analysis import RawSpikeAnalysis36 37 38logging.basicConfig(level=logging.DEBUG)39logger = logging.getLogger(__name__)40 41# Initialize Panel with Bootstrap and Tabulator extensions42pn.extension("tabulator")43curdoc().title = "LC-NE Patch-seq Data Explorer"44 45 46def add_efel_asymmetry_columns(df_meta: pd.DataFrame) -> None:47 rise_cols = {}48 fall_cols = {}49 50 for col in df_meta.columns:51 if not isinstance(col, str):52 continue53 if not col.startswith("efel_") or " @ " not in col:54 continue55 metric, suffix = col.split(" @ ", 1)56 match = re.match(r"^(efel_.+?)_(rise|fall)(.*)$", metric)57 if not match:58 continue59 base = f"{match.group(1)}{match.group(3)}"60 key = f"{base} @ {suffix}"61 if match.group(2) == "rise":62 rise_cols[key] = col63 else:64 fall_cols[key] = col65 66 for key in sorted(set(rise_cols) & set(fall_cols)):67 base, suffix = key.split(" @ ", 1)68 new_col = f"{base}_asymmetry @ {suffix}"69 rise_vals = pd.to_numeric(df_meta[rise_cols[key]], errors="coerce")70 fall_vals = pd.to_numeric(df_meta[fall_cols[key]], errors="coerce")71 df_meta[new_col] = np.where(fall_vals != 0, rise_vals / fall_vals, np.nan)72 73 74class PatchSeqNWBApp(param.Parameterized):75 """76 Object-Oriented Panel App for navigating NWB files.77 Encapsulates metadata loading, sweep visualization, and cell selection.78 """79 80 class DataHolder(param.Parameterized):81 """82 Holder for currently selected cell ID and sweep number.83 """84 85 ephys_roi_id_selected = param.String(default="")86 sweep_number_selected = param.Integer(default=0)87 filtered_df_meta = param.DataFrame()88 89 def __init__(self):90 """91 Initialize the PatchSeqNWBApp.92 """93 # Holder for currently selected cell ID.94 self.data_holder = PatchSeqNWBApp.DataHolder()95 96 # Store figures and state for context-aware export97 self._cell_explorer_figures = {}98 self._active_tab = 099 self._export_progress = pn.widgets.Progress(100 name="Export progress",101 value=0,102 max=100,103 sizing_mode="stretch_width",104 bar_color="success",105 active=True,106 visible=False,107 )108 109 # Load and prepare metadata.110 self.df_meta = load_ephys_metadata(111 if_from_s3=True, if_with_seq=True, if_with_morphology=True112 )113 add_efel_asymmetry_columns(self.df_meta)114 self.df_meta.rename(115 columns={116 "x": "X (A --> P)",117 "y": "Y (D --> V)",118 "z": "Z (L --> R)",119 },120 inplace=True,121 )122 123 # Preprocess data: Convert NaN in "virus" column to "None"124 if "virus" in self.df_meta.columns:125 self.df_meta["virus"] = self.df_meta["virus"].fillna("None")126 127 self.cell_key = [128 "Date",129 "jem-id_cell_specimen",130 "ephys_roi_id",131 "ephys_qc",132 "LC_targeting",133 "injection region",134 "Y (D --> V)",135 ]136 # Turn Date to datetime137 self.df_meta.loc[:, "Date_str"] = self.df_meta[138 "Date"139 ] # Keep the original Date as string140 self.df_meta.loc[:, "Date"] = pd.to_datetime(141 self.df_meta["Date"], errors="coerce"142 )143 144 # Initialize scatter plot component145 self.scatter_plot = ScatterPlot(self.df_meta, self.data_holder)146 147 # Initialize spike analysis component148 self.raw_spike_analysis = RawSpikeAnalysis(self.df_meta, main_app=self)149 150 # Create a copy for filtering - this will be updated by the global filter151 self.data_holder.filtered_df_meta = self.df_meta.copy()152 153 def update_bokeh(self, raw, sweep, downsample_factor=3):154 """155 Update the Bokeh plot for a given sweep.156 """157 trace = raw.get_raw_trace(sweep)[::downsample_factor]158 stimulus = raw.get_stimulus(sweep)[::downsample_factor]159 time = raw.get_time(sweep)[::downsample_factor]160 161 box_zoom_auto = BoxZoomTool(dimensions="auto")162 163 # Create the voltage trace plot164 voltage_plot = figure(165 title=f"Full traces - Sweep number {sweep} (downsampled {downsample_factor}x)",166 height=300,167 tools=["hover", box_zoom_auto, "box_zoom", "wheel_zoom", "reset", "pan"],168 active_drag=box_zoom_auto,169 x_range=(0, time[-1]),170 y_axis_label="Vm (mV)",171 sizing_mode="stretch_width",172 )173 voltage_plot.line(time, trace, line_width=1.5, color="navy")174 175 # Create the stimulus plot176 stim_plot = figure(177 height=150,178 tools=["hover", box_zoom_auto, "box_zoom", "wheel_zoom", "reset", "pan"],179 active_drag=box_zoom_auto,180 x_range=voltage_plot.x_range, # Link x ranges181 x_axis_label="Time (ms)",182 y_axis_label="I (pA)",183 sizing_mode="stretch_width",184 )185 stim_plot.line(time, stimulus, line_width=1.5, color="firebrick")186 187 # Store figures for export188 self._cell_explorer_figures = {189 "voltage_trace": voltage_plot,190 "stimulus_trace": stim_plot,191 }192 193 # Stack the plots vertically using bokeh's column layout194 layout = bokeh_column(195 voltage_plot, stim_plot, sizing_mode="stretch_width", margin=(50, 0, 0, 0)196 )197 return layout198 199 @staticmethod200 def highlight_selected_rows(row, highlight_subset, color, fields=None):201 """202 Highlight rows based on a subset of values.203 If fields is None, highlight the entire row.204 """205 style = [""] * len(row)206 if row["sweep_number"] in highlight_subset:207 if fields is None:208 return [f"background-color: {color}"] * len(row)209 else:210 for field in fields:211 style[list(row.keys()).index(field)] = f"background-color: {color}"212 return style213 214 @staticmethod215 def get_qc_message(sweep, df_sweeps):216 """Return a QC message based on sweep data."""217 if sweep not in df_sweeps["sweep_number"].values:218 return "<span style='color:red;'>Invalid sweep!</span>"219 if sweep in df_sweeps.query("passed != passed")["sweep_number"].values:220 return "<span style='background:salmon;'>Sweep terminated by the experimenter!</span>"221 if sweep in df_sweeps.query("passed == False")["sweep_number"].values:222 return (223 f"<span style='background:yellow;'>Sweep failed QC! "224 f"({df_sweeps[df_sweeps.sweep_number == sweep].reasons.iloc[0][0]})</span>"225 )226 return "<span style='background:lightgreen;'>Sweep passed QC!</span>"227 228 def _download_svg_context_aware(self):229 """Download SVGs based on the active tab."""230 from components.utils.svg_export import export_figures_to_svg_zip231 232 progress = self._export_progress233 progress.value = 0234 progress.visible = True235 progress.name = "Preparing export..."236 237 try:238 if self._active_tab == 0: # Cell Explorer tab239 figures = {}240 progress.name = "Collecting scatter figures..."241 242 if (243 hasattr(self.scatter_plot, "_latest_figures")244 and self.scatter_plot._latest_figures245 ):246 figures.update(self.scatter_plot._latest_figures)247 248 progress.name = "Collecting cell explorer figures..."249 250 if self._cell_explorer_figures:251 figures.update(self._cell_explorer_figures)252 253 if not figures:254 raise RuntimeError(255 "No figures available. Select a cell and generate scatter plot first."256 )257 258 prefix = "app_download_cell_explorer"259 elif self._active_tab == 1: # Spike Analysis tab260 progress.name = "Collecting spike analysis figures..."261 262 if not self.raw_spike_analysis._latest_figures:263 raise RuntimeError(264 "Generate the spike analysis plots before downloading SVGs."265 )266 figures = self.raw_spike_analysis._latest_figures267 prefix = "app_download_spike_analysis"268 else: # Other tabs269 raise RuntimeError(270 f"SVG export not supported for tab {self._active_tab}."271 )272 273 exportable_count = sum(1 for fig in figures.values() if fig is not None)274 if exportable_count == 0:275 raise RuntimeError(276 "No exportable figures available for the current tab."277 )278 279 progress.value = 0280 progress.name = f"Rendering SVGs (0/{exportable_count})"281 282 def report_progress(current, total):283 progress.value = min(100, int((current / total) * 100))284 progress.name = f"Rendering SVGs ({current}/{total})"285 286 zip_buffer, timestamp = export_figures_to_svg_zip(287 figures, progress_callback=report_progress288 )289 290 progress.value = 100291 progress.name = "Packing download..."292 self._download_button.filename = f"{prefix}_{timestamp}.zip"293 progress.name = "Ready"294 295 return zip_buffer296 finally:297 progress.visible = False298 progress.value = 0299 progress.name = "Export progress"300 301 def apply_global_filter(self, query_string):302 """303 Apply a query filter to the metadata DataFrame.304 305 Args:306 query_string: A string in pandas query format to filter the metadata307 308 Returns:309 Filtered DataFrame310 """311 if not query_string.strip():312 # If query is empty, reset to the full dataset313 self.data_holder.filtered_df_meta = self.df_meta.copy()314 return f"Reset to full dataset (N={len(self.data_holder.filtered_df_meta)})"315 316 try:317 # Apply the filter query318 filtered = self.df_meta.query(query_string)319 if len(filtered) == 0:320 return "Query returned 0 results. Filter not applied."321 322 # Update the filtered dataframe in the data_holder323 # This will trigger updates in any components bound to this parameter324 self.data_holder.filtered_df_meta = filtered325 326 return f"Query applied. {len(filtered)} records match (out of {len(self.df_meta)})."327 except Exception as e:328 return f"Error in query: {str(e)}"329 330 def create_scatter_plot(self):331 """332 Create the scatter plot panel using the ScatterPlot component.333 """334 335 # Get plot controls from the scatter plot component336 controls = self.scatter_plot.controls337 control_width = 300338 339 # Create a reactive scatter plot that updates when controls change340 scatter_plot = pn.bind(341 self.scatter_plot.update_scatter_plot,342 controls["x_axis_select"].param.value,343 controls["y_axis_select"].param.value,344 controls["color_col_select"].param.value,345 controls["color_palette_select"].param.value,346 controls["size_col_select"].param.value,347 controls["size_range_slider"].param.value,348 controls["size_gamma_slider"].param.value,349 controls["alpha_slider"].param.value,350 controls["width_slider"].param.value,351 controls["height_slider"].param.value,352 controls["font_size_slider"].param.value,353 controls["bins_slider"].param.value,354 controls["hist_height_slider"].param.value,355 controls["show_gmm"].param.value,356 controls["n_components_x"].param.value,357 controls["n_components_y"].param.value,358 controls["show_linear_fit"].param.value,359 df_meta=self.data_holder.param.filtered_df_meta,360 )361 362 return pn.Row(363 pn.Column(364 controls["x_axis_select"],365 controls["y_axis_select"],366 pn.layout.Divider(margin=(5, 0, 5, 0)),367 controls["color_col_select"],368 controls["color_palette_select"],369 pn.layout.Divider(margin=(5, 0, 5, 0)),370 controls["size_col_select"],371 controls["size_range_slider"],372 controls["size_gamma_slider"],373 pn.layout.Divider(margin=(5, 0, 5, 0)),374 controls["bins_slider"],375 controls["show_gmm"],376 controls["n_components_x"],377 controls["n_components_y"],378 controls["show_linear_fit"],379 pn.layout.Divider(margin=(5, 0, 5, 0)),380 pn.Accordion(381 (382 "Plot settings",383 pn.Column(384 controls["alpha_slider"],385 controls["width_slider"],386 controls["height_slider"],387 controls["hist_height_slider"],388 controls["font_size_slider"],389 width=control_width - 30,390 ),391 ),392 active=[1],393 ),394 margin=(0, 50, 20, 0), # top, right, bottom, left margins in pixels395 width=control_width,396 ),397 scatter_plot,398 margin=(0, 20, 20, 20), # top, right, bottom, left margins in pixels399 # width=800,400 )401 402 def _sync_url_state(self, tabs, spike_controls, filter_query=None):403 """Centralize all URL sync logic for the app."""404 405 location = pn.state.location406 location.sync(tabs, {"active": "tab"})407 location.sync(408 self.data_holder,409 {410 "ephys_roi_id_selected": "cell_id",411 "sweep_number_selected": "sweep",412 },413 )414 415 spike_mapping = {416 "extract_from": "spike_extract",417 "spike_type": "spike_type",418 "dim_reduction_method": "dim_method",419 "spike_range": "spike_range",420 "normalize_window_v": "norm_v",421 "normalize_window_dvdt": "norm_dvdt",422 "n_clusters": "n_clusters",423 "if_show_cluster_on_retro": "show_retro",424 "marker_size": "marker_size",425 "alpha_slider": "alpha",426 "plot_width": "plot_width",427 "plot_height": "plot_height",428 "font_size": "font_size",429 }430 for control_name, url_param in spike_mapping.items():431 location.sync(spike_controls[control_name], {"value": url_param})432 433 self.scatter_plot.sync_controls_to_url()434 435 if filter_query is not None:436 location.sync(filter_query, {"value": "query"})437 438 def create_cell_selector_panel(self, filtered_df_meta):439 """440 Builds and returns the cell selector panel that displays metadata.441 """442 # MultiSelect widget to choose additional columns.443 cols = list(filtered_df_meta.columns)444 cols.sort()445 selectable_cols = [col for col in cols if col not in self.cell_key]446 col_selector = pn.widgets.MultiSelect(447 name="Add more columns to show in the table",448 options=selectable_cols,449 value=[450 "ipfx_width_rheo",451 "efel_AP_width @ long_square_rheo, aver",452 "ipfx_sag",453 "efel_sag_ratio1 @ subthreshold, aver",454 ], # start with no additional columns455 height=300,456 width=500,457 )458 459 def add_df_meta_col(selected_columns, filtered_df_meta):460 """461 Add selected columns from the filtered metadata DataFrame.462 """463 return filtered_df_meta[self.cell_key + selected_columns]464 465 filtered_df_meta = pn.bind(add_df_meta_col, col_selector, filtered_df_meta)466 tab_df_meta = pn.widgets.Tabulator(467 filtered_df_meta,468 selectable=1,469 disabled=True, # Not editable470 frozen_columns=self.cell_key,471 groupby=["injection region"],472 header_filters=True,473 show_index=False,474 height=300,475 sizing_mode="stretch_width",476 pagination=None,477 stylesheets=[":host .tabulator {font-size: 12px;}"],478 )479 480 # When a row is selected, update the current cell (ephys_roi_id).481 def update_sweep_view_from_table(event):482 if event.new:483 selected_index = event.new[0]484 self.data_holder.ephys_roi_id_selected = str(485 int(486 self.data_holder.filtered_df_meta.iloc[selected_index][487 "ephys_roi_id"488 ]489 )490 )491 492 tab_df_meta.param.watch(update_sweep_view_from_table, "selection")493 494 scatter_plot = self.create_scatter_plot()495 496 cell_selector_panel = pn.Column(497 pn.Row(498 col_selector,499 tab_df_meta,500 height=350,501 ),502 pn.Row(503 scatter_plot,504 ),505 )506 return cell_selector_panel507 508 def create_sweep_panel(self, ephys_roi_id=""):509 """510 Builds and returns the sweep visualization panel for a single cell.511 """512 if ephys_roi_id == "":513 return pn.pane.Markdown("Please select a cell from the table above.")514 515 # Load the NWB file for the selected cell.516 raw_this_cell = PatchSeqNWB(ephys_roi_id=ephys_roi_id, if_load_metadata=False)517 518 # Now let's get df sweep from the eFEL enriched one519 df_sweeps = load_efel_features_from_roi(ephys_roi_id, if_from_s3=True)[520 "df_sweeps"521 ]522 df_sweeps_valid = df_sweeps.query("passed == passed")523 524 # Set initial sweep number to first valid sweep525 if self.data_holder.sweep_number_selected == 0:526 self.data_holder.sweep_number_selected = int(527 df_sweeps_valid.iloc[0]["sweep_number"]528 )529 530 # Add a slider to control the downsample factor531 downsample_factor = pn.widgets.IntSlider(532 name="Downsample factor",533 value=5,534 start=1,535 end=10,536 )537 538 # Bind the plotting function to the data holder's sweep number539 bokeh_panel = pn.bind(540 self.update_bokeh,541 raw=raw_this_cell,542 sweep=self.data_holder.param.sweep_number_selected,543 downsample_factor=downsample_factor.param.value_throttled,544 )545 546 # Bind the S3 URL retrieval to the data holder's sweep number547 def get_s3_sweep_images(sweep_number):548 s3_url = get_public_url_sweep(ephys_roi_id, sweep_number)549 images = []550 if isinstance(s3_url, dict) and "sweep" in s3_url:551 images.append(pn.pane.PNG(s3_url["sweep"], width=800, height=400))552 if isinstance(s3_url, dict) and "spikes" in s3_url:553 images.append(pn.pane.PNG(s3_url["spikes"], width=800, height=400))554 return (555 pn.Column(*images)556 if images557 else pn.pane.Markdown("No S3 images available")558 )559 560 s3_sweep_images_panel = pn.bind(561 get_s3_sweep_images,562 sweep_number=self.data_holder.param.sweep_number_selected,563 )564 sweep_pane = pn.Column(565 s3_sweep_images_panel,566 bokeh_panel,567 downsample_factor,568 sizing_mode="stretch_width",569 )570 571 # Build a Tabulator for sweep metadata.572 tab_sweeps = pn.widgets.Tabulator(573 df_sweeps_valid[574 [575 "sweep_number",576 "stimulus_code_ext",577 "stimulus_name",578 "stimulus_amplitude",579 "passed",580 "efel_num_spikes",581 "num_spikes",582 "stimulus_start_time",583 "stimulus_duration",584 "tags",585 "reasons",586 "stimulus_code",587 ]588 ], # Only show valid sweeps (passed is not NaN)589 hidden_columns=["stimulus_code"],590 selectable=1,591 disabled=True, # Not editable592 frozen_columns=["sweep_number"],593 header_filters=True,594 show_index=False,595 height=700,596 width=1000,597 groupby=["stimulus_code"],598 stylesheets=[":host .tabulator {font-size: 12px;}"],599 )600 601 # Apply conditional row highlighting.602 if hasattr(tab_sweeps, "style"):603 tab_sweeps.style.apply(604 PatchSeqNWBApp.highlight_selected_rows,605 highlight_subset=df_sweeps_valid.query("passed == True")[606 "sweep_number"607 ].tolist(),608 color="lightgreen",609 fields=["passed"],610 axis=1,611 ).apply(612 PatchSeqNWBApp.highlight_selected_rows,613 highlight_subset=df_sweeps_valid.query("passed != passed")[614 "sweep_number"615 ].tolist(),616 color="salmon",617 fields=["passed"],618 axis=1,619 ).apply(620 PatchSeqNWBApp.highlight_selected_rows,621 highlight_subset=df_sweeps_valid.query("passed == False")[622 "sweep_number"623 ].tolist(),624 color="yellow",625 fields=["passed"],626 axis=1,627 ).apply(628 PatchSeqNWBApp.highlight_selected_rows,629 highlight_subset=df_sweeps_valid.query("num_spikes > 0")[630 "sweep_number"631 ].tolist(),632 color="lightgreen",633 fields=["num_spikes"],634 axis=1,635 )636 637 # --- Synchronize table selection with sweep number ---638 def update_sweep_from_table(event):639 """Update sweep number when table selection changes."""640 if event.new:641 selected_index = event.new[0]642 new_sweep = int(df_sweeps_valid.iloc[selected_index]["sweep_number"])643 self.data_holder.sweep_number_selected = new_sweep644 645 tab_sweeps.param.watch(update_sweep_from_table, "selection")646 # --- End Synchronization ---647 648 # Build a reactive QC message panel.649 sweep_msg = pn.bind(650 PatchSeqNWBApp.get_qc_message,651 sweep=self.data_holder.param.sweep_number_selected,652 df_sweeps=df_sweeps,653 )654 sweep_msg_panel = pn.pane.Markdown(sweep_msg, width=600, height=30)655 656 return pn.Row(657 pn.Column(658 pn.pane.Markdown(f"# {ephys_roi_id}"),659 pn.pane.Markdown("Select a sweep from the table to view its data."),660 pn.Column(sweep_msg_panel, sweep_pane),661 width=700,662 margin=(0, 100, 0, 0), # top, right, bottom, left margins663 ),664 pn.Column(665 pn.pane.Markdown("## Sweep metadata"),666 tab_sweeps,667 ),668 )669 670 def main_layout(self):671 """672 Constructs the full application layout with Bootstrap template.673 """674 pn.config.throttled = False675 676 # Create and bind the cell selector panel to filtered metadata677 pane_cell_selector = pn.bind(678 self.create_cell_selector_panel,679 filtered_df_meta=self.data_holder.param.filtered_df_meta,680 )681 682 # Create spike analysis controls and plots683 spike_controls = self.raw_spike_analysis.create_plot_controls()684 685 def update_spike_plots(686 extract_from,687 spike_type,688 n_clusters,689 alpha,690 width,691 height,692 marker_size,693 if_show_cluster_on_retro,694 if_edge_color_projection,695 normalize_window_v,696 normalize_window_dvdt,697 spike_range,698 dim_reduction_method,699 font_size,700 filtered_df_meta=None,701 ):702 df_spikes = self.raw_spike_analysis.get_spikes(spike_type)703 # Extract representative spikes (normalized) - with peak alignment for time-series plot704 df_v_norm, df_dvdt_norm = extract_representative_spikes(705 df_spikes=df_spikes,706 extract_from=extract_from,707 if_normalize_v=True,708 normalize_window_v=normalize_window_v,709 if_normalize_dvdt=True,710 normalize_window_dvdt=normalize_window_dvdt,711 if_smooth_dvdt=False,712 if_align_dvdt_peaks=True,713 filtered_df_meta=filtered_df_meta,714 )715 716 # Extract representative spikes (normalized) - without peak alignment for normalized phase plot717 df_v_norm_phase, df_dvdt_norm_phase = extract_representative_spikes(718 df_spikes=df_spikes,719 extract_from=extract_from,720 if_normalize_v=True,721 normalize_window_v=normalize_window_v,722 if_normalize_dvdt=True,723 normalize_window_dvdt=normalize_window_dvdt,724 if_smooth_dvdt=False,725 if_align_dvdt_peaks=False,726 filtered_df_meta=filtered_df_meta,727 )728 729 # Extract representative spikes (unnormalized) - without peak alignment for phase plots730 df_v_unnorm, df_dvdt_unnorm = extract_representative_spikes(731 df_spikes=df_spikes,732 extract_from=extract_from,733 if_normalize_v=False,734 normalize_window_v=normalize_window_v,735 if_normalize_dvdt=False,736 normalize_window_dvdt=normalize_window_dvdt,737 if_smooth_dvdt=False,738 if_align_dvdt_peaks=False,739 filtered_df_meta=filtered_df_meta,740 )741 742 # Create spike analysis plots with the filtered dataframe743 return self.raw_spike_analysis.create_raw_PCA_plots(744 df_v_norm=df_v_norm,745 df_dvdt_norm=df_dvdt_norm,746 df_v_phase_norm=df_v_norm_phase,747 df_dvdt_phase_norm=df_dvdt_norm_phase,748 df_v_unnorm=df_v_unnorm,749 df_dvdt_unnorm=df_dvdt_unnorm,750 n_clusters=n_clusters,751 alpha=alpha,752 width=width,753 height=height,754 marker_size=marker_size,755 if_show_cluster_on_retro=if_show_cluster_on_retro,756 if_edge_color_projection=if_edge_color_projection,757 spike_range=spike_range,758 dim_reduction_method=dim_reduction_method,759 font_size=font_size,760 normalize_window_v=normalize_window_v,761 normalize_window_dvdt=normalize_window_dvdt,762 )763 764 # Create spike analysis plots765 controls = spike_controls # shorter name for readability766 param_keys = [767 "n_clusters",768 "alpha_slider",769 "plot_width",770 "if_show_cluster_on_retro",771 "if_edge_color_projection",772 "plot_height",773 "marker_size",774 "normalize_window_v",775 "normalize_window_dvdt",776 "spike_range",777 "dim_reduction_method",778 "font_size",779 ]780 params = {781 k: (782 controls[k].param.value783 if k not in ["if_show_cluster_on_retro", "dim_reduction_method"]784 else controls[k].param.value785 )786 for k in param_keys787 }788 789 spike_plots = pn.bind(790 update_spike_plots,791 extract_from=controls["extract_from"].param.value,792 spike_type=controls["spike_type"].param.value,793 n_clusters=params["n_clusters"],794 alpha=params["alpha_slider"],795 width=params["plot_width"],796 height=params["plot_height"],797 marker_size=params["marker_size"],798 if_show_cluster_on_retro=params["if_show_cluster_on_retro"],799 if_edge_color_projection=params["if_edge_color_projection"],800 normalize_window_v=params["normalize_window_v"],801 normalize_window_dvdt=params["normalize_window_dvdt"],802 spike_range=params["spike_range"],803 dim_reduction_method=params["dim_reduction_method"],804 font_size=params["font_size"],805 filtered_df_meta=self.data_holder.param.filtered_df_meta,806 )807 808 # Create cell summary plot809 def get_s3_cell_summary_plot(ephys_roi_id):810 s3_url = get_public_url_cell_summary(ephys_roi_id)811 if s3_url:812 return pn.pane.PNG(s3_url, sizing_mode="stretch_width")813 else:814 return pn.pane.Markdown(815 "### Select the table or the scatter plot to view the cell summary plot."816 )817 818 s3_cell_summary_plot = pn.Column(819 pn.bind(820 lambda ephys_roi_id: pn.pane.Markdown(821 "## Cell summary plot"822 + (f" for {ephys_roi_id}" if ephys_roi_id else "")823 ),824 ephys_roi_id=self.data_holder.param.ephys_roi_id_selected,825 ),826 pn.bind(827 get_s3_cell_summary_plot,828 ephys_roi_id=self.data_holder.param.ephys_roi_id_selected,829 ),830 sizing_mode="stretch_width",831 )832 833 # Bind the sweep panel to the current cell selection.834 pane_one_cell = pn.bind(835 self.create_sweep_panel,836 ephys_roi_id=self.data_holder.param.ephys_roi_id_selected,837 )838 839 # Create a toggle button for showing/hiding raw sweeps840 show_sweeps_button = pn.widgets.Button(841 name="Show raw sweeps", button_type="primary", width=200842 )843 show_sweeps = pn.widgets.Toggle(name="Show raw sweeps", value=False)844 845 # Link the button to the toggle846 def toggle_sweeps(event):847 show_sweeps.value = not show_sweeps.value848 show_sweeps_button.name = (849 "Hide raw sweeps" if show_sweeps.value else "Show raw sweeps"850 )851 852 show_sweeps_button.on_click(toggle_sweeps)853 854 # Create a dynamic layout that includes pane_one_cell only when show_sweeps is True855 dynamic_content = pn.bind(856 lambda show: pn.Column(pane_one_cell) if show else pn.Column(),857 show_sweeps.param.value,858 )859 860 # --- Connect global filter components ---861 filter_query = pn.widgets.TextAreaInput(862 name="Query string",863 value="`jem-status_reporter` == 'Positive' & `injection region` not in ['Non-Retro', 'Thalamus']",864 placeholder="Enter a pandas query string",865 sizing_mode="stretch_width",866 height=100,867 )868 869 filter_button = pn.widgets.Button(870 name="Apply filter",871 button_type="primary",872 width=150,873 )874 875 reset_button = pn.widgets.Button(876 name="Reset filter",877 button_type="light",878 width=150,879 )880 881 filter_status = pn.pane.Markdown("", css_classes=["alert", "p-2", "m-2"])882 883 # Connect the button to the filter function884 def apply_filter_callback(event):885 result = self.apply_global_filter(filter_query.value)886 if "reset" in result.lower() or "success" in result.lower():887 filter_status.css_classes = ["alert", "alert-success", "p-2", "m-2"]888 elif "error" in result.lower():889 filter_status.css_classes = ["alert", "alert-danger", "p-2", "m-2"]890 else:891 filter_status.css_classes = ["alert", "alert-info", "p-2", "m-2"]892 filter_status.object = result893 894 def reset_filter_callback(event):895 filter_query.value = ""896 result = self.apply_global_filter("")897 filter_status.css_classes = ["alert", "alert-success", "p-2", "m-2"]898 filter_status.object = result899 900 filter_button.on_click(apply_filter_callback)901 reset_button.on_click(reset_filter_callback)902 # --- End filter components ---903 904 # Build the filter panel905 filter_panel = pn.Column(906 pn.pane.Markdown("### Global Filter", css_classes=["card-title"]),907 pn.pane.Markdown(908 """909 #### Enter a pandas query to filter cells. Examples:910 LC targeting911 - (Only retro cells) `` `injection region` != 'Non-Retro' ``912 913 Gene QC914 - (Nucleus present) `` `jem-nucleus_post_patch` == "nucleus_present" ``915 - (QC based on mitochondrial RNA) `` `gene_RNA_QC (log_normed)` == True ``916 917 Gene expression918 - (Fluorescence+) `` `jem-status_reporter` == "Positive" ``919 - (Dbh subclass) `` mapmycells_subclass_name.str.contains("DBH", case=False, na=False) ``920 - (Marker genes any positive) `` `gene_Dbh (log_normed)` > 0 or `gene_Th (log_normed)` > 0 or921 `gene_Slc18a2 (log_normed)` > 0 or `gene_Slc6a2 (log_normed)` > 0 ``922 - (Dbh+) `` `gene_Dbh (log_normed)` > 0 ``923 924 Location925 - `` `X (A --> P)` > 9500 and `X (A --> P)` < 11500 and926 `Y (D --> V)` > 2500 and `Y (D --> V)` < 6000 ``927 """928 ),929 pn.Column(930 filter_query,931 pn.Row(filter_button, reset_button),932 ),933 filter_status,934 width=800,935 margin=(0, 100, 50, 0), # top, right, bottom, left margins in pixels936 css_classes=["card", "p-4", "m-4"],937 )938 939 if not hasattr(self, "_filter_autoload_registered"):940 941 def _apply_filter_on_load():942 apply_filter_callback(None)943 944 pn.state.onload(_apply_filter_on_load)945 self._filter_autoload_registered = True946 947 # Create the filtered count display948 filtered_count = pn.bind(949 lambda filtered_df: pn.pane.Markdown(950 f"### Filtered cells: {len(filtered_df)} of {len(self.df_meta)} total",951 css_classes=["alert", "alert-info", "p-2", "text-center"],952 ),953 filtered_df=self.data_holder.param.filtered_df_meta,954 )955 956 # Create tabs for different sections957 tabs = pn.Tabs(958 (959 "Cell Explorer",960 pn.Column(961 filtered_count,962 pane_cell_selector,963 ),964 ),965 (966 "Spike Analysis",967 pn.Card(968 pn.Row(969 pn.Column(970 pn.pane.Markdown(971 "### Controls", css_classes=["card-title"]972 ),973 *spike_controls.values(),974 width=250,975 ),976 pn.Column(spike_plots),977 ),978 title="Raw Spike Analysis",979 collapsed=False,980 ),981 ),982 (983 "Raw Sweeps",984 pn.Column(985 show_sweeps_button,986 pn.pane.Markdown(987 "### Select a cell from the Cell Explorer tab to view its raw sweeps",988 css_classes=["alert", "alert-info", "p-2"],989 ),990 dynamic_content,991 ),992 ),993 (994 "Feature Distribution",995 pn.Card(996 pn.pane.PNG(997 S3_PUBLIC_URL_BASE998 + "/efel/cell_stats/distribution_all_features.png",999 width=1300,1000 ),1001 title="Distribution of all features",1002 collapsed=False,1003 ),1004 ),1005 dynamic=True, # Allow dynamic updates to tab content1006 )1007 1008 self._sync_url_state(tabs, spike_controls, filter_query)1009 1010 # Initialize _active_tab from current tab state (important for URL loading)1011 self._active_tab = tabs.active1012 1013 # Create download button in sidebar1014 self._download_button = pn.widgets.FileDownload(1015 label="Download figures (SVG)",1016 filename="plots.zip",1017 button_type="success",1018 sizing_mode="stretch_width",1019 )1020 self._download_button.callback = self._download_svg_context_aware1021 1022 # Track active tab to change download behavior1023 def update_active_tab(event):1024 self._active_tab = event.new1025 1026 tabs.param.watch(update_active_tab, "active")1027 1028 # Create the template1029 template = pn.template.BootstrapTemplate(1030 title="LC-NE Patch-seq Data Explorer",1031 header_background="#0072B5", # Allen Institute blue1032 favicon=(1033 "https://alleninstitute.org/wp-content/uploads/2021/10/"1034 "cropped-favicon-32x32.png"1035 ),1036 main=[1037 # pn.pane.Markdown("# Patch-seq Ephys Data Explorer", css_classes=["display-4"]),1038 # pn.layout.Divider(),1039 pn.Row(filter_panel, s3_cell_summary_plot),1040 tabs,1041 ],1042 sidebar=[1043 pn.pane.Markdown("### Filtered Cells"),1044 pn.bind(1045 lambda filtered_df: pn.pane.Markdown(1046 f"**{len(filtered_df)} of {len(self.df_meta)} total**",1047 css_classes=["alert", "alert-info", "p-2"],1048 ),1049 filtered_df=self.data_holder.param.filtered_df_meta,1050 ),1051 pn.pane.Markdown("### Selected Cell"),1052 pn.bind(1053 lambda id: pn.pane.Markdown(1054 f"**Cell ID:** {id}" if id else "No cell selected",1055 css_classes=["alert", "alert-secondary", "p-2"],1056 ),1057 id=self.data_holder.param.ephys_roi_id_selected,1058 ),1059 pn.bind(1060 lambda id: pn.pane.Markdown(1061 f"**Sweep:** {id}" if id else "",1062 css_classes=["alert", "alert-secondary", "p-2"],1063 ),1064 id=self.data_holder.param.sweep_number_selected,1065 ),1066 pn.layout.Divider(margin=(10, 0)),1067 pn.pane.Markdown("### Export"),1068 self._download_button,1069 self._export_progress,1070 ],1071 theme="default",1072 )1073 template.sidebar_width = 2001074 return template1075 1076 1077app = PatchSeqNWBApp()1078layout = app.main_layout()1079layout.servable()1080