workbench 0.8.224__py3-none-any.whl → 0.8.234__py3-none-any.whl

workbench/web_interface/components/plugins/model_details.py

@@ -3,14 +3,19 @@
 from typing import Union
 
 # Dash Imports
-from dash import html, callback, dcc, Input, Output, State
+from dash import html, callback, dcc, no_update, Input, Output, State
 
 # Workbench Imports
 from workbench.api import ModelType, ParameterStore
 from workbench.cached.cached_model import CachedModel
-from workbench.utils.markdown_utils import health_tag_markdown
+from workbench.utils.markdown_utils import (
+    health_tag_markdown,
+    tags_to_markdown,
+    dict_to_markdown,
+    dict_to_collapsible_html,
+    df_to_html_table,
+)
 from workbench.web_interface.components.plugin_interface import PluginInterface, PluginPage, PluginInputType
-from workbench.utils.markdown_utils import tags_to_markdown, dict_to_markdown, dict_to_collapsible_html
 
 
 class ModelDetails(PluginInterface):
@@ -44,7 +49,7 @@ class ModelDetails(PluginInterface):
                 dcc.Markdown(id=f"{self.component_id}-summary", dangerously_allow_html=True),
                 html.H5(children="Inference Metrics", style={"marginTop": "20px"}),
                 dcc.Dropdown(id=f"{self.component_id}-dropdown", className="dropdown"),
-                dcc.Markdown(id=f"{self.component_id}-metrics"),
+                dcc.Markdown(id=f"{self.component_id}-metrics", dangerously_allow_html=True),
             ],
         )
 
@@ -66,7 +71,9 @@ class ModelDetails(PluginInterface):
 
         Args:
             model (CachedModel): An instantiated CachedModel object
-            **kwargs: Additional keyword arguments (unused)
+            **kwargs: Additional keyword arguments
+                - inference_run: Current inference run selection (to preserve user's choice)
+                - previous_model_name: Name of the previously selected model
 
         Returns:
             list: A list of the updated property values for the plugin
@@ -80,10 +87,21 @@ class ModelDetails(PluginInterface):
 
         # Populate the inference runs dropdown
         inference_runs, default_run = self.get_inference_runs()
-        metrics = self.inference_metrics(default_run)
 
-        # Return the updated property values for the plugin
-        return [header, details, inference_runs, default_run, metrics]
+        # Check if the model changed
+        previous_model_name = kwargs.get("previous_model_name")
+        current_inference_run = kwargs.get("inference_run")
+        model_changed = previous_model_name != model.name
+
+        # Only preserve the inference run if the model hasn't changed AND the selection is valid
+        if not model_changed and current_inference_run and current_inference_run in inference_runs:
+            # Same model, preserve the user's selection - use no_update for dropdown value
+            metrics = self.inference_metrics(current_inference_run)
+            return [header, details, inference_runs, no_update, metrics]
+        else:
+            # New model or invalid selection - use default
+            metrics = self.inference_metrics(default_run)
+            return [header, details, inference_runs, default_run, metrics]
 
     def register_internal_callbacks(self):
         @callback(
@@ -175,15 +193,14 @@ class ModelDetails(PluginInterface):
             markdown += "  \nNo Data  \n"
         else:
             markdown += "  \n"
-            metrics = metrics.round(3)
 
             # If the model is a classification model, have the index sorting match the class labels
             if self.current_model.model_type == ModelType.CLASSIFIER:
-                # Sort the metrics by the class labels (if they match)
                 class_labels = self.current_model.class_labels()
                 if set(metrics.index) == set(class_labels):
                     metrics = metrics.reindex(class_labels)
-            markdown += metrics.to_markdown()
+
+            markdown += df_to_html_table(metrics)
 
         # Get additional inference metrics if they exist
         model_name = self.current_model.name

workbench/web_interface/components/plugins/scatter_plot.py

@@ -1,6 +1,7 @@
 import base64
+import numpy as np
 import pandas as pd
-from dash import dcc, html, callback, Input, Output, no_update
+from dash import dcc, html, callback, clientside_callback, Input, Output, no_update
 import plotly.graph_objects as go
 import plotly.express as px
 from dash.exceptions import PreventUpdate
@@ -8,7 +9,9 @@ from dash.exceptions import PreventUpdate
 # Workbench Imports
 from workbench.web_interface.components.plugin_interface import PluginInterface, PluginPage, PluginInputType
 from workbench.utils.theme_manager import ThemeManager
-from workbench.utils.plot_utils import prediction_intervals, molecule_hover_tooltip
+from workbench.utils.plot_utils import prediction_intervals
+from workbench.utils.chem_utils.vis import molecule_hover_tooltip
+from workbench.utils.clientside_callbacks import circle_overlay_callback
 
 
 class ScatterPlot(PluginInterface):
@@ -35,10 +38,10 @@ class ScatterPlot(PluginInterface):
         self.df = None
         self.show_axes = show_axes
         self.theme_manager = ThemeManager()
-        self.colorscale = self.theme_manager.colorscale()
         self.has_smiles = False  # Track if dataframe has smiles column for molecule hover
         self.smiles_column = None
         self.id_column = None
+        self.hover_background = None  # Cached background color for molecule hover tooltip
 
         # Call the parent class constructor
         super().__init__()
@@ -78,7 +81,7 @@ class ScatterPlot(PluginInterface):
                     id=f"{component_id}-graph",
                     figure=self.display_text("Waiting for Data..."),
                     config={"scrollZoom": True},
-                    style={"height": "100%"},
+                    style={"height": "500px", "width": "100%"},
                     clear_on_unhover=True,
                 ),
                 # Controls: X, Y, Color, Label Dropdowns, and Regression Line Checkbox
@@ -178,6 +181,10 @@ class ScatterPlot(PluginInterface):
             list: A list of updated property values (figure, x options, y options, color options,
                                                     x default, y default, color default).
         """
+        # Get the colorscale and background color from the current theme
+        self.colorscale = self.theme_manager.colorscale()
+        self.hover_background = self.theme_manager.background()
+
         # Get the limit for the number of rows to plot
         limit = kwargs.get("limit", 20000)
 
@@ -198,7 +205,10 @@ class ScatterPlot(PluginInterface):
 
         # Check if the dataframe has smiles/id columns for molecule hover rendering
         self.smiles_column = next((col for col in self.df.columns if col.lower() == "smiles"), None)
-        self.id_column = kwargs.get("id_column") or next((col for col in self.df.columns if col.lower() == "id"), None)
+        # Use provided id_column, or auto-detect "id" column, or fall back to first column
+        self.id_column = kwargs.get("id_column") or next(
+            (col for col in self.df.columns if col.lower() == "id"), self.df.columns[0]
+        )
         self.has_smiles = self.smiles_column is not None
 
         # Identify numeric columns
@@ -206,10 +216,11 @@ class ScatterPlot(PluginInterface):
         if len(numeric_columns) < 3:
             raise ValueError("At least three numeric columns are required for x, y, and color.")
 
-        # Default x, y, and color (for color, default to a numeric column)
+        # Default x, y, and color (for color, prefer 'confidence' if it exists)
         x_default = kwargs.get("x", numeric_columns[0])
         y_default = kwargs.get("y", numeric_columns[1])
-        color_default = kwargs.get("color", numeric_columns[2])
+        default_color = "confidence" if "confidence" in self.df.columns else numeric_columns[2]
+        color_default = kwargs.get("color", default_color)
         regression_line = kwargs.get("regression_line", False)
 
         # Create the default scatter plot
@@ -238,7 +249,6 @@ class ScatterPlot(PluginInterface):
         y_col: str,
         color_col: str,
         regression_line: bool = False,
-        marker_size: int = 15,
     ) -> go.Figure:
         """Create a Plotly Scatter Plot figure.
 
@@ -248,12 +258,20 @@ class ScatterPlot(PluginInterface):
             y_col (str): The column to use for the y-axis.
             color_col (str): The column to use for the color scale.
             regression_line (bool): Whether to include a regression line.
-            marker_size (int): Size of the markers. Default is 15.
 
         Returns:
             go.Figure: A Plotly Figure object.
         """
 
+        # If aggregation_count is present, sort so largest counts are drawn first (underneath)
+        # and compute marker sizes using square root (between log and linear)
+        if "aggregation_count" in df.columns:
+            df = df.sort_values("aggregation_count", ascending=False).reset_index(drop=True)
+            # Scale: base_size (15) + (sqrt(count) - 1) * factor, so count=1 stays at base_size
+            marker_sizes = 15 + (np.sqrt(df["aggregation_count"]) - 1) * 3
+        else:
+            marker_sizes = 15
+
         # Helper to generate hover text for each point.
         def generate_hover_text(row):
             return "<br>".join([f"{col}: {row[col]}" for col in self.hover_columns])
@@ -295,11 +313,11 @@ class ScatterPlot(PluginInterface):
                     hovertemplate=hovertemplate,
                     customdata=df[custom_data_cols] if custom_data_cols else None,
                     marker=dict(
-                        size=marker_size,
+                        size=marker_sizes,
                         color=marker_color,
                         colorscale=self.colorscale,
                         colorbar=colorbar,
-                        opacity=0.8,
+                        opacity=0.9,
                         line=dict(color="rgba(0,0,0,0.25)", width=1),
                     ),
                 )
@@ -315,6 +333,15 @@ class ScatterPlot(PluginInterface):
             for i, cat in enumerate(categories):
                 sub_df = df[df[color_col] == cat]
                 sub_hovertext = hovertext.loc[sub_df.index] if hovertext is not None else None
+                # Get marker sizes for this subset (handles both array and scalar)
+                if isinstance(marker_sizes, (pd.Series, np.ndarray)):
+                    sub_marker_sizes = (
+                        marker_sizes.loc[sub_df.index]
+                        if isinstance(marker_sizes, pd.Series)
+                        else marker_sizes[sub_df.index]
+                    )
+                else:
+                    sub_marker_sizes = marker_sizes
                 trace = go.Scattergl(
                     x=sub_df[x_col],
                     y=sub_df[y_col],
@@ -325,7 +352,7 @@ class ScatterPlot(PluginInterface):
                     hovertemplate=hovertemplate,
                     customdata=sub_df[custom_data_cols] if custom_data_cols else None,
                     marker=dict(
-                        size=marker_size,
+                        size=sub_marker_sizes,
                         color=discrete_colors[i % len(discrete_colors)],
                         opacity=0.8,
                         line=dict(color="rgba(0,0,0,0.25)", width=1),
@@ -405,35 +432,14 @@ class ScatterPlot(PluginInterface):
 
             raise PreventUpdate
 
-        @callback(
+        # Clientside callback for circle overlay - runs in browser, no server round trip
+        clientside_callback(
+            circle_overlay_callback(self._circle_data_uri),
             Output(f"{self.component_id}-overlay", "show"),
             Output(f"{self.component_id}-overlay", "bbox"),
             Output(f"{self.component_id}-overlay", "children"),
             Input(f"{self.component_id}-graph", "hoverData"),
         )
-        def _scatter_circle_overlay(hover_data):
-            """Show white circle overlay centered on the hovered point."""
-            if hover_data is None:
-                return False, no_update, no_update
-
-            # Extract bounding box from hoverData
-            bbox = hover_data["points"][0]["bbox"]
-
-            # Use pre-computed circle SVG
-            svg_image = html.Img(src=self._circle_data_uri, style={"width": "100px", "height": "100px"})
-
-            # Get the center of the bounding box
-            center_x = (bbox["x0"] + bbox["x1"]) / 2
-            center_y = (bbox["y0"] + bbox["y1"]) / 2
-
-            # The tooltip should be centered on the point
-            adjusted_bbox = {
-                "x0": center_x - 50,
-                "x1": center_x + 50,
-                "y0": center_y - 162,
-                "y1": center_y - 62,
-            }
-            return True, adjusted_bbox, [svg_image]
 
         @callback(
             Output(f"{self.component_id}-molecule-tooltip", "show"),
@@ -459,17 +465,24 @@ class ScatterPlot(PluginInterface):
                 smiles = customdata
                 mol_id = None
 
-            # Generate molecule tooltip with ID header
+            # Generate molecule tooltip with ID header (use cached background color)
             mol_width, mol_height = 300, 200
-            children = molecule_hover_tooltip(smiles, mol_id=mol_id, width=mol_width, height=mol_height)
+            children = molecule_hover_tooltip(
+                smiles, mol_id=mol_id, width=mol_width, height=mol_height, background=self.hover_background
+            )
 
-            # Extract bounding box and offset the molecule tooltip to the right of the point
+            # Position molecule tooltip above and slightly right of the point
             bbox = hover_data["points"][0]["bbox"]
+            center_x = (bbox["x0"] + bbox["x1"]) / 2
+            center_y = (bbox["y0"] + bbox["y1"]) / 2
+            x_offset = 5  # Slight offset to the right
+            y_offset = mol_height + 50  # Above the point
+
             adjusted_bbox = {
-                "x0": bbox["x0"] + 15,
-                "x1": bbox["x1"] + mol_width + 15,
-                "y0": bbox["y0"] - (2 * mol_height + 60),
-                "y1": bbox["y1"] - (mol_height + 60),
+                "x0": center_x + x_offset,
+                "x1": center_x + x_offset + mol_width,
+                "y0": center_y - mol_height - y_offset,
+                "y1": center_y - y_offset,
             }
             return True, adjusted_bbox, children
 

workbench/web_interface/components/settings_menu.py

@@ -75,8 +75,9 @@ class SettingsMenu:
             style={"display": "flex", "flexDirection": "column", "alignItems": "center", "justifyContent": "center"},
         )
 
-        # Build menu items: Status, License, divider, Themes submenu
+        # Build menu items: Home, Status, License, divider, Themes submenu
         menu_items = [
+            dbc.DropdownMenuItem("Home", href="/"),
             dbc.DropdownMenuItem("Status", href="/status", external_link=True, target="_blank"),
             dbc.DropdownMenuItem("License", href="/license", external_link=True, target="_blank"),
             dbc.DropdownMenuItem(divider=True),

workbench/web_interface/page_views/main_page.py

@@ -3,7 +3,6 @@
 import pandas as pd
 from typing import Optional, Tuple
 
-
 # Workbench Imports
 from workbench.web_interface.page_views.page_view import PageView
 from workbench.utils.symbols import tag_symbols

{workbench-0.8.224.dist-info → workbench-0.8.234.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: workbench
-Version: 0.8.224
+Version: 0.8.234
 Summary: Workbench: A Dashboard and Python API for creating and deploying AWS SageMaker Model Pipelines
 Author-email: SuperCowPowers LLC <support@supercowpowers.com>
 License: MIT License
@@ -60,7 +60,7 @@ Requires-Dist: plotly>=6.0.0; extra == "ui"
 Requires-Dist: dash>=3.0.0; extra == "ui"
 Requires-Dist: dash-bootstrap-components>=1.6.0; extra == "ui"
 Requires-Dist: dash-bootstrap-templates>=1.3.0; extra == "ui"
-Requires-Dist: dash_ag_grid; extra == "ui"
+Requires-Dist: dash_ag_grid>=33.3.3; extra == "ui"
 Requires-Dist: tabulate>=0.9.0; extra == "ui"
 Requires-Dist: matplotlib>=3.9.2; extra == "ui"
 Provides-Extra: dev
@@ -77,51 +77,49 @@ Requires-Dist: workbench[dev]; extra == "all"
 Dynamic: license-file
 
 
-# Recent News
+## Live Dashboard Demo
+You can explore a live demo of the Workbench Dashboard at: [Workbench Dashboard Demo](https://workbench-dashboard.com)
 
-### Themes
+## Recent News
+**Chemprop Models!** All the rage for the Open ADMET Challenge. 
 
-Everyone knows that good data science requires... **Some Awesome Themes!**
+Workbench now supports:
+- Single Task Chemprop Models
+- Multi Task Chemprop Models
+- Chemprop Hybrid Models (MPNN + Descriptors)
+- Foundation Chemprop Models (CheMeleon Pretrained)
+
+Examples: 
+
+- [Deploying Chemprop Models](examples/models/chemprop.py)
+- [Deploying Foundation Chemprop Models](examples/models/chemprop_foundation.py)
+
+**References**
+- [Open ADMET Challenge](https://huggingface.co/spaces/openadmet/OpenADMET-ExpansionRx-Challenge)
+- **ChemProp:** Yang et al. "Analyzing Learned Molecular Representations for Property Prediction" *J. Chem. Inf. Model.* 2019 — [GitHub](https://github.com/chemprop/chemprop) | [Paper](https://pubs.acs.org/doi/10.1021/acs.jcim.9b00237)
+- [CheMeleon Github](https://github.com/JacksonBurns/chemeleon)
+
+### Chemprop Action Shots!
 
 <table>
   <tr>
     <td>
-      <a href="https://github.com/user-attachments/assets/82ab4eab-0688-4b93-ad8e-9b954564777b">
-        <img width="400" alt="theme_dark" src="https://github.com/user-attachments/assets/82ab4eab-0688-4b93-ad8e-9b954564777b" />
-      </a>
-    </td>
-    <td>
-      <a href="https://github.com/user-attachments/assets/b63a0789-c144-4048-afb6-f03e3d993680">
-        <img width="400" alt="theme_light" src="https://github.com/user-attachments/assets/b63a0789-c144-4048-afb6-f03e3d993680" />
+      <a href="https://github.com/user-attachments/assets/a36c6eff-c464-4c9a-9859-a45cd7e35145">
+        <img width="800" alt="theme_dark" src="https://github.com/user-attachments/assets/a36c6eff-c464-4c9a-9859-a45cd7e35145" />
       </a>
     </td>
   </tr>
   <tr>
     <td>
-      <a href="https://github.com/user-attachments/assets/8a59be19-0c5d-42c6-9922-feafb1a1eecd">
-        <img width="400" alt="theme_quartz" src="https://github.com/user-attachments/assets/8a59be19-0c5d-42c6-9922-feafb1a1eecd" />
-      </a>
-    </td>
-    <td>
-      <a href="https://github.com/user-attachments/assets/5b01ec64-8d56-43bf-96c5-7da8ec48f527">
-        <img width="400" alt="theme_quartz_dark" src="https://github.com/user-attachments/assets/5b01ec64-8d56-43bf-96c5-7da8ec48f527" />
+      <a href="https://github.com/user-attachments/assets/d65ec1da-e04e-44fe-8782-4da0fb50588a">
+        <img width="800" alt="theme_quartz" src="https://github.com/user-attachments/assets/d65ec1da-e04e-44fe-8782-4da0fb50588a" />
       </a>
     </td>
   </tr>
 </table>
 
-All of the Dashboard pages, subpages, and plugins use our new `ThemeManager()` class. See [Workbench Themes](https://supercowpowers.github.io/workbench/themes/), also big thanks to our friends at [Dash Bootstrap Templates](https://github.com/AnnMarieW/dash-bootstrap-templates)
-
 
 
-### Workbench up on the AWS Marketplace
-
-Powered by AWS® to accelerate your Machine Learning Pipelines development with our new [Dashboard for ML Pipelines](https://aws.amazon.com/marketplace/pp/prodview-5idedc7uptbqo). Getting started with Workbench is a snap and can be billed through AWS.
-
-### Coming Soon: `v0.9`
-
-We're getting ready for our `v0.9` release. Here's the road map: [Workbench RoadMaps](https://supercowpowers.github.io/workbench/road_maps/0_9_0/) 
-
 # Welcome to Workbench
 The Workbench framework makes AWS® both easier to use and more powerful. Workbench handles all the details around updating and managing a complex set of AWS Services. With a simple-to-use Python API and a beautiful set of web interfaces, Workbench makes creating AWS ML pipelines a snap. It also dramatically improves both the usability and visibility across the entire spectrum of services: Glue Job, Athena, Feature Store, Models, and Endpoints, Workbench makes it easy to build production ready, AWS powered, machine learning pipelines.
 
@@ -159,6 +157,10 @@ For the full instructions for connecting your AWS Account see:
 - One time AWS Onboarding: [AWS Setup](https://supercowpowers.github.io/workbench/aws_setup/core_stack/)
 
 
+### Workbench up on the AWS Marketplace
+
+Powered by AWS® to accelerate your Machine Learning Pipelines development with our new [Dashboard for ML Pipelines](https://aws.amazon.com/marketplace/pp/prodview-5idedc7uptbqo). Getting started with Workbench is a snap and can be billed through AWS.
+
 ### Workbench Presentations
 Even though Workbench makes AWS easier, it's taking something very complex (the full set of AWS ML Pipelines/Services) and making it less complex. Workbench has a depth and breadth of functionality so we've provided higher level conceptual documentation See: [Workbench Presentations](https://supercowpowers.github.io/workbench/presentations/)