Merge pull request #237 from BiAPoL/histogram1d

Implement Histogram1D if the same measurement is selected for both axes

README.md

@@ -95,9 +95,11 @@ on `Update Axes/Clustering Options` to refresh them. Click on `Plot` to draw the
 
 Currently, you can select between two types of plots: scatter plot and 2D histogram. Click on "expand for advanced
 options" to see the selection of the plot type. Clustering (manual and automatic) is possible using both types of plots.
-2D histogram is recommended if you have a very high number of data points. Under advanced options you will also find the
-checkbox determining whether not selected data points should be automatically clustered as another cluster or displayed as
-gray data points and not visualized in the generated clusters IDs layer.
+2D histogram is recommended if you have a very high number of data points. If you select the same measurement for both axes,
+and histogram is selected as the plotting type, 1D histogram will be plotted. Clustering 1D histogram is not yet possible.
+
+Under advanced options you will also find the checkbox determining whether not-selected data points should be automatically
+clustered as another cluster or displayed as gray data points and not visualized in the generated clusters IDs layer.
 
 ![](https://github.com/BiAPoL/napari-clusters-plotter/raw/main/images/plot_plain.png)
 

napari_clusters_plotter/_Qt_code.py

@@ -10,7 +10,7 @@
 from matplotlib.backends.backend_qt5agg import NavigationToolbar2QT as NavigationToolbar
 from matplotlib.figure import Figure
 from matplotlib.path import Path
-from matplotlib.widgets import LassoSelector, RectangleSelector
+from matplotlib.widgets import LassoSelector, RectangleSelector, SpanSelector
 from napari.layers import Image, Labels
 from qtpy.QtCore import QRect
 from qtpy.QtGui import QIcon
@@ -414,6 +414,43 @@ def disconnect(self):
         self.canvas.draw_idle()
 
 
+class SelectFrom1DHistogram:
+    def __init__(self, parent, ax, full_data):
+        self.parent = parent
+        self.ax = ax
+        self.canvas = ax.figure.canvas
+        self.xys = full_data
+
+        self.span_selector = SpanSelector(
+            ax,
+            onselect=self.onselect,
+            direction="horizontal",
+            props=dict(facecolor="#1f77b4", alpha=0.5),
+        )
+        self.click_id = self.canvas.mpl_connect("button_press_event", self.on_click)
+
+    def onselect(self, vmin, vmax):
+        self.ind_mask = np.logical_and(self.xys >= vmin, self.xys <= vmax).values
+
+        if self.parent.manual_clustering_method is not None:
+            self.parent.manual_clustering_method(self.ind_mask)
+
+    def on_click(self, event):
+        # Clear selection if user right-clicks (without moving) outside of the histogram
+        if event.inaxes != self.ax:
+            return
+        if event.button == 3:
+            # clear selection
+            self.ind_mask = np.zeros_like(self.xys, dtype=bool)
+            if self.parent.manual_clustering_method is not None:
+                self.parent.manual_clustering_method(self.ind_mask)
+
+    def disconnect(self):
+        self.span_selector.disconnect_events()
+        self.canvas.mpl_disconnect(self.click_id)
+        self.canvas.draw_idle()
+
+
 # Class below was based upon matplotlib lasso selection example:
 # https://matplotlib.org/stable/gallery/widgets/lasso_selector_demo_sgskip.html
 class SelectFromCollection:
@@ -564,11 +601,42 @@ def make_2d_histogram(
         self.axes.set_ylim(yedges[0], yedges[-1])
         self.histogram = (h, xedges, yedges)
 
-        full_data = pd.concat([data_x, data_y], axis=1)
+        full_data = pd.concat([pd.DataFrame(data_x), pd.DataFrame(data_y)], axis=1)
         self.selector.disconnect()
         self.selector = SelectFrom2DHistogram(self, self.axes, full_data)
         self.axes.figure.canvas.draw_idle()
 
+    def make_1d_histogram(
+        self,
+        data: "numpy.typing.ArrayLike",
+        bin_number: int = 400,
+        log_scale: bool = False,
+    ):
+        counts, bins = np.histogram(data, bins=bin_number)
+        self.axes.hist(
+            bins[:-1],
+            bins,
+            edgecolor="white",
+            weights=counts,
+            log=log_scale,
+            color="#9A9A9A",
+        )
+        self.histogram = (counts, bins)
+        bin_width = bins[1] - bins[0]
+        self.axes.set_xlim(min(bins) - (bin_width / 2), max(bins) + (bin_width / 2))
+        ymin = 0
+        if log_scale:
+            ymin = 1
+        self.axes.set_ylim(ymin, max(counts) * 1.1)
+
+        if log_scale:
+            self.axes.set_xscale("linear")
+            self.axes.set_yscale("log")
+
+        self.selector.disconnect()
+        self.selector = SelectFrom1DHistogram(self, self.axes, data)
+        self.axes.figure.canvas.draw_idle()
+
     def make_scatter_plot(
         self,
         data_x: "numpy.typing.ArrayLike",
@@ -613,6 +681,7 @@ def match_napari_layout(self):
         # changing colors of axes labels
         self.axes.xaxis.label.set_color("white")
         self.axes.yaxis.label.set_color("white")
+        self.fig.canvas.draw_idle()
 
 
 # overriding NavigationToolbar method to change the background and axes colors of saved figure

napari_clusters_plotter/_plotter.py

@@ -24,6 +24,7 @@
 )
 
 from ._plotter_utilities import (
+    apply_cluster_colors_to_bars,
     clustered_plot_parameters,
     estimate_number_bins,
     make_cluster_overlay_img,
@@ -53,7 +54,7 @@
 
 
 class PlottingType(Enum):
-    HISTOGRAM_2D = auto()
+    HISTOGRAM = auto()
     SCATTER = auto()
 
 
@@ -193,7 +194,6 @@ def replot():
             features = get_layer_tabular_data(self.analysed_layer)
 
             # redraw the whole plot
-
             try:
                 self.run(
                     features,
@@ -210,7 +210,7 @@ def checkbox_status_changed():
             replot()
 
         def plotting_type_changed():
-            if self.plotting_type.currentText() == PlottingType.HISTOGRAM_2D.name:
+            if self.plotting_type.currentText() == PlottingType.HISTOGRAM.name:
                 self.bin_number_container.setVisible(True)
                 self.log_scale_container.setVisible(True)
                 self.plot_hide_non_selected.setChecked(True)
@@ -235,7 +235,7 @@ def bin_auto():
         combobox_plotting_container.layout().addWidget(QLabel("Plotting type"))
         self.plotting_type = QComboBox()
         self.plotting_type.addItems(
-            [PlottingType.SCATTER.name, PlottingType.HISTOGRAM_2D.name]
+            [PlottingType.SCATTER.name, PlottingType.HISTOGRAM.name]
         )
         self.plotting_type.currentIndexChanged.connect(plotting_type_changed)
         combobox_plotting_container.layout().addWidget(self.plotting_type)
@@ -427,7 +427,10 @@ def reset_choices(self, event=None):
     def change_state_of_nonselected_checkbox(self):
         # make the checkbox visible only if clustering is done manually
         visible = (
-            True if "MANUAL_CLUSTER_ID" in self.plot_cluster_id.currentText() else False
+            True
+            if "MANUAL_CLUSTER_ID" in self.plot_cluster_id.currentText()
+            or self.plot_cluster_id.currentText() == ""
+            else False
         )
         self.hide_nonselected_checkbox_container.setVisible(visible)
 
@@ -556,51 +559,77 @@ def run(
                 self.graphics_widget.make_scatter_plot(
                     self.data_x, self.data_y, colors_plot, sizes, a
                 )
+
+                self.graphics_widget.axes.set_xlabel(plot_x_axis_name)
+                self.graphics_widget.axes.set_ylabel(plot_y_axis_name)
             else:
                 if self.bin_auto.isChecked():
-                    number_bins = int(
-                        np.max(
-                            [
-                                estimate_number_bins(self.data_x),
-                                estimate_number_bins(self.data_y),
-                            ]
+                    if plot_x_axis_name == plot_y_axis_name:
+                        number_bins = int(estimate_number_bins(self.data_x))
+                    else:
+                        number_bins = int(
+                            np.max(
+                                [
+                                    estimate_number_bins(self.data_x),
+                                    estimate_number_bins(self.data_y),
+                                ]
+                            )
                         )
-                    )
-                    self.bin_number_spinner.setValue(number_bins)
+                        self.bin_number_spinner.setValue(number_bins)
                 else:
                     number_bins = int(self.bin_number_spinner.value())
 
-                self.graphics_widget.make_2d_histogram(
-                    self.data_x,
-                    self.data_y,
-                    colors,
-                    bin_number=number_bins,
-                    log_scale=self.log_scale.isChecked(),
-                )
+                # if both axes are the same, plot 1D histogram
+                if plot_x_axis_name == plot_y_axis_name:
+                    self.graphics_widget.make_1d_histogram(
+                        self.data_x,
+                        bin_number=number_bins,
+                        log_scale=self.log_scale.isChecked(),
+                    )
+                    # update bar colors to cluster ids
+                    self.graphics_widget.axes = apply_cluster_colors_to_bars(
+                        self.graphics_widget.axes,
+                        cluster_name=plot_cluster_name,
+                        features=features,
+                        number_bins=number_bins,
+                        feature_x=self.plot_x_axis_name,
+                        colors=colors,
+                    )
+                    self.graphics_widget.figure.canvas.draw_idle()
+                    self.graphics_widget.axes.set_xlabel(plot_x_axis_name)
+                    self.graphics_widget.axes.set_ylabel("frequency")
+                else:
+                    self.graphics_widget.make_2d_histogram(
+                        self.data_x,
+                        self.data_y,
+                        colors,
+                        bin_number=number_bins,
+                        log_scale=self.log_scale.isChecked(),
+                    )
 
-                rgb_img = make_cluster_overlay_img(
-                    cluster_id=plot_cluster_name,
-                    features=features,
-                    feature_x=self.plot_x_axis_name,
-                    feature_y=self.plot_y_axis_name,
-                    colors=colors,
-                    histogram_data=self.graphics_widget.histogram,
-                    hide_first_cluster=self.plot_hide_non_selected.isChecked(),
-                )
-                xedges = self.graphics_widget.histogram[1]
-                yedges = self.graphics_widget.histogram[2]
-
-                self.graphics_widget.axes.imshow(
-                    rgb_img,
-                    extent=[xedges[0], xedges[-1], yedges[0], yedges[-1]],
-                    origin="lower",
-                    alpha=1,
-                    aspect="auto",
-                )
-                self.graphics_widget.figure.canvas.draw_idle()
+                    rgb_img = make_cluster_overlay_img(
+                        cluster_id=plot_cluster_name,
+                        features=features,
+                        feature_x=self.plot_x_axis_name,
+                        feature_y=self.plot_y_axis_name,
+                        colors=colors,
+                        histogram_data=self.graphics_widget.histogram,
+                        hide_first_cluster=self.plot_hide_non_selected.isChecked(),
+                    )
+                    xedges = self.graphics_widget.histogram[1]
+                    yedges = self.graphics_widget.histogram[2]
+
+                    self.graphics_widget.axes.imshow(
+                        rgb_img,
+                        extent=[xedges[0], xedges[-1], yedges[0], yedges[-1]],
+                        origin="lower",
+                        alpha=1,
+                        aspect="auto",
+                    )
+                    self.graphics_widget.figure.canvas.draw_idle()
+                    self.graphics_widget.axes.set_xlabel(plot_x_axis_name)
+                    self.graphics_widget.axes.set_ylabel(plot_y_axis_name)
 
-            self.graphics_widget.axes.set_xlabel(plot_x_axis_name)
-            self.graphics_widget.axes.set_ylabel(plot_y_axis_name)
             self.graphics_widget.match_napari_layout()
 
             from vispy.color import Color
@@ -705,31 +734,47 @@ def run(
                 self.graphics_widget.make_scatter_plot(
                     self.data_x, self.data_y, colors_plot, sizes, a
                 )
+
+                self.graphics_widget.axes.set_xlabel(plot_x_axis_name)
+                self.graphics_widget.axes.set_ylabel(plot_y_axis_name)
             else:
                 if self.bin_auto.isChecked():
-                    number_bins = int(
-                        np.max(
-                            [
-                                estimate_number_bins(self.data_x),
-                                estimate_number_bins(self.data_y),
-                            ]
+                    if plot_x_axis_name == plot_y_axis_name:
+                        number_bins = int(estimate_number_bins(self.data_x))
+                    else:
+                        number_bins = int(
+                            np.max(
+                                [
+                                    estimate_number_bins(self.data_x),
+                                    estimate_number_bins(self.data_y),
+                                ]
+                            )
                         )
-                    )
                     self.bin_number_spinner.setValue(number_bins)
                 else:
                     number_bins = int(self.bin_number_spinner.value())
 
-                self.graphics_widget.make_2d_histogram(
-                    self.data_x,
-                    self.data_y,
-                    colors,
-                    bin_number=number_bins,
-                    log_scale=self.log_scale.isChecked(),
-                )
-            self.graphics_widget.axes.set_xlabel(plot_x_axis_name)
-            self.graphics_widget.axes.set_ylabel(plot_y_axis_name)
+                # if both axes are the same, plot 1D histogram
+                if plot_x_axis_name == plot_y_axis_name:
+                    self.graphics_widget.make_1d_histogram(
+                        self.data_x,
+                        bin_number=number_bins,
+                        log_scale=self.log_scale.isChecked(),
+                    )
+                    self.graphics_widget.axes.set_xlabel(plot_x_axis_name)
+                    self.graphics_widget.axes.set_ylabel("frequency")
+                else:
+                    self.graphics_widget.make_2d_histogram(
+                        self.data_x,
+                        self.data_y,
+                        colors,
+                        bin_number=number_bins,
+                        log_scale=self.log_scale.isChecked(),
+                    )
+                    self.graphics_widget.axes.set_xlabel(plot_x_axis_name)
+                    self.graphics_widget.axes.set_ylabel(plot_y_axis_name)
+
             self.graphics_widget.match_napari_layout()
 
-            self.graphics_widget.draw()  # Only redraws when cluster is not manually selected
-            # because manual selection already does that when selector is disconnected
+        self.graphics_widget.draw()  # Always redraws, oterwise y-axis may not get updated in histograms
         self.graphics_widget.reset_zoom()