initial feature scatter

Author: kevinyamauchi

src/napari_matplotlib/base.py

@@ -79,13 +79,14 @@ def setup_callbacks(self) -> None:
         # z-step changed in viewer
         self.viewer.dims.events.current_step.connect(self._draw)
         # Layer selection changed in viewer
-        self.viewer.layers.selection.events.active.connect(self.update_layers)
+        self.viewer.layers.selection.events.changed.connect(self.update_layers)
 
     def update_layers(self, event: napari.utils.events.Event) -> None:
         """
         Update the currently selected layers and re-draw.
         """
         self.layers = list(self.viewer.layers.selection)
+        self._on_update_layers()
         self._draw()
 
     def _draw(self) -> None:
@@ -95,6 +96,7 @@ def _draw(self) -> None:
         """
         self.clear()
         if self.n_selected_layers != self.n_layers_input:
+            self.canvas.draw()
             return
         self.draw()
         self.canvas.draw()
@@ -112,3 +114,9 @@ def draw(self) -> None:
 
         This is a no-op, and is intended for derived classes to override.
         """
+
+    def _on_update_layers(self) -> None:
+        """This function is called when self.layers is updated via self.update_layer()
+
+        This is a no-op, and is intended for derived classes to override.
+        """

src/napari_matplotlib/scatter.py

@@ -1,39 +1,188 @@
+from typing import List, Tuple, Union
+
 import matplotlib.colors as mcolor
 import napari
+import numpy as np
+from magicgui import magicgui
 
 from .base import NapariMPLWidget
 
 __all__ = ["ScatterWidget"]
 
 
-class ScatterWidget(NapariMPLWidget):
-    """
-    Widget to display scatter plot of two similarly shaped layers.
+class ScatterBaseWidget(NapariMPLWidget):
+    def __init__(
+        self,
+        napari_viewer: napari.viewer.Viewer,
+        marker_alpha: float = 0.5,
+        histogram_for_large_data: bool = True,
+    ):
+        super().__init__(napari_viewer)
 
-    If there are more than 500 data points, a 2D histogram is displayed instead
-    of a scatter plot, to avoid too many scatter points.
-    """
+        # flag set to True if histogram should be used
+        # for plotting large points
+        self.histogram_for_large_data = histogram_for_large_data
 
-    n_layers_input = 2
+        # set plotting visualization attributes
+        self._marker_alpha = 0.5
 
-    def __init__(self, napari_viewer: napari.viewer.Viewer):
-        super().__init__(napari_viewer)
         self.axes = self.canvas.figure.subplots()
         self.update_layers(None)
 
+    @property
+    def marker_alpha(self) -> float:
+        """Alpha (opacity) for the scatter markers"""
+        return self._marker_alpha
+
+    @marker_alpha.setter
+    def marker_alpha(self, alpha: float):
+        self._marker_alpha = alpha
+        self._draw()
+
+    def clear(self) -> None:
+        self.axes.clear()
+
     def draw(self) -> None:
         """
         Clear the axes and scatter the currently selected layers.
         """
-        data = [layer.data[self.current_z] for layer in self.layers]
-        if data[0].size < 500:
-            self.axes.scatter(data[0], data[1], alpha=0.5)
-        else:
+        data, x_axis_name, y_axis_name = self._get_data()
+
+        if len(data) == 0:
+            # don't plot if there isn't data
+            return
+
+        if self.histogram_for_large_data and (data[0].size > 500):
             self.axes.hist2d(
                 data[0].ravel(),
                 data[1].ravel(),
                 bins=100,
                 norm=mcolor.LogNorm(),
             )
-        self.axes.set_xlabel(self.layers[0].name)
-        self.axes.set_ylabel(self.layers[1].name)
+        else:
+            self.axes.scatter(data[0], data[1], alpha=self.marker_alpha)
+
+        self.axes.set_xlabel(x_axis_name)
+        self.axes.set_ylabel(y_axis_name)
+
+    def _get_data(self) -> Tuple[np.ndarray, str, str]:
+        raise NotImplementedError
+
+
+class ScatterWidget(ScatterBaseWidget):
+    """
+    Widget to display scatter plot of two similarly shaped layers.
+
+    If there are more than 500 data points, a 2D histogram is displayed instead
+    of a scatter plot, to avoid too many scatter points.
+    """
+
+    n_layers_input = 2
+
+    def __init__(
+        self,
+        napari_viewer: napari.viewer.Viewer,
+        marker_alpha: float = 0.5,
+        histogram_for_large_data: bool = True,
+    ):
+        super().__init__(
+            napari_viewer,
+            marker_alpha=marker_alpha,
+            histogram_for_large_data=histogram_for_large_data,
+        )
+
+    def _get_data(self) -> Tuple[np.ndarray, str, str]:
+        data = [layer.data[self.current_z] for layer in self.layers]
+        x_axis_name = self.layers[0].name
+        y_axis_name = self.layers[1].name
+
+        return data, x_axis_name, y_axis_name
+
+
+class FeaturesScatterWidget(ScatterBaseWidget):
+    n_layers_input = 1
+
+    def __init__(
+        self,
+        napari_viewer: napari.viewer.Viewer,
+        marker_alpha: float = 0.5,
+        histogram_for_large_data: bool = True,
+        key_selection_gui: bool = True,
+    ):
+        self._key_selection_widget = None
+        super().__init__(
+            napari_viewer,
+            marker_alpha=marker_alpha,
+            histogram_for_large_data=histogram_for_large_data,
+        )
+
+        if key_selection_gui is True:
+            self._key_selection_widget = magicgui(
+                self._set_axis_keys,
+                x_axis_key={"choices": self._get_valid_axis_keys},
+                y_axis_key={"choices": self._get_valid_axis_keys},
+                call_button="plot",
+            )
+            self.layout().addWidget(self._key_selection_widget.native)
+
+    @property
+    def x_axis_key(self) -> Union[None, str]:
+        """Key to access x axis data from the FeaturesTable"""
+        return self._x_axis_key
+
+    @x_axis_key.setter
+    def x_axis_key(self, key: Union[None, str]):
+        self._x_axis_key = key
+        self._draw()
+
+    @property
+    def y_axis_key(self) -> Union[None, str]:
+        """Key to access y axis data from the FeaturesTable"""
+        return self._y_axis_key
+
+    @y_axis_key.setter
+    def y_axis_key(self, key: Union[None, str]):
+        self._y_axis_key = key
+        self._draw()
+
+    def _set_axis_keys(self, x_axis_key: str, y_axis_key: str):
+        """Set both axis keys and then redraw the plot"""
+        self._x_axis_key = x_axis_key
+        self._y_axis_key = y_axis_key
+        self._draw()
+
+    def _get_valid_axis_keys(self, combo_widget=None) -> List[str]:
+        if len(self.layers) == 0:
+            return []
+        else:
+            return self.layers[0].features.keys()
+
+    def _get_data(self) -> Tuple[np.ndarray, str, str]:
+        feature_table = self.layers[0].features
+
+        if (
+            (len(feature_table) == 0)
+            or (self.x_axis_key is None)
+            or (self.y_axis_key is None)
+        ):
+            return np.array([]), "", ""
+
+        data_x = feature_table[self.x_axis_key]
+        data_y = feature_table[self.y_axis_key]
+        data = np.stack((data_x, data_y))
+
+        x_axis_name = self.x_axis_key.replace("_", " ")
+        y_axis_name = self.y_axis_key.replace("_", " ")
+
+        return data, x_axis_name, y_axis_name
+
+    def _on_update_layers(self) -> None:
+        """This is called when the layer selection changes
+        by self.update_layers().
+        """
+        if self._key_selection_widget is not None:
+            self._key_selection_widget.reset_choices()
+
+        # reset the axis keys
+        self._x_axis_key = None
+        self._y_axis_key = None