Unify figure style

Author: schilling40

scripts/figures/plot_fig2.py

@@ -135,81 +135,73 @@ def fig_02c(save_path, plot=False, all_versions=False):
     ihc_annotator = [0.958, 0.956, 0.957]
     syn_unet = [0.931, 0.905, 0.918]
 
-    # This is the version with IHC v4b segmentation:
-    # 4th version of the network with optimized segmentation params
-    version_1 = [sgn_unet, sgn_annotator, ihc_v4b, ihc_annotator, syn_unet]
-    settings_1 = ["automatic", "manual", "automatic", "manual", "automatic"]
+    setting = ["SGN", "IHC", "Synapse"]
 
     # This is the version with IHC v4c segmentation:
     # 4th version of the network with optimized segmentation params and split of falsely merged IHCs
-    version_2 = [sgn_unet, sgn_annotator, ihc_v4c, ihc_annotator, syn_unet]
-    settings_2 = ["automatic", "manual", "automatic", "manual", "automatic"]
+    manual = [sgn_annotator, ihc_annotator]
+    automatic = [sgn_unet, ihc_v4c, syn_unet]
 
-    # This is the version with IHC v4c + filter segmentation:
-    # 4th version of the network with optimized segmentation params and split of falsely merged IHCs
-    # + filtering out IHCs with zero mapped synapses.
-    version_3 = [sgn_unet, sgn_annotator, ihc_v4c_filter, ihc_annotator, syn_unet]
-    settings_3 = ["automatic", "manual", "automatic", "manual", "automatic"]
+    precision_manual = [i[0] for i in manual]
+    recall_manual = [i[1] for i in manual]
+    f1score_manual = [i[2] for i in manual]
 
-    if all_versions:
-        versions = [version_1, version_2, version_3]
-        settings = [settings_1, settings_2, settings_3]
-        save_suffix = ["_v4b", "_v4c", "_v4c_filter"]
-        save_paths = [save_path.split(".")[0] + i + "." + save_path.split(".")[1] for i in save_suffix]
-    else:
-        versions = [version_2]
-        settings = [settings_2]
-        save_suffix = ["_v4c"]
-        save_paths = [save_path.split(".")[0] + i + "." + save_path.split(".")[1] for i in save_suffix]
+    precision_automatic = [i[0] for i in automatic]
+    recall_automatic = [i[1] for i in automatic]
+    f1score_automatic = [i[2] for i in automatic]
 
-    for version, setting, save_path in zip(versions, settings, save_paths):
-        precision = [i[0] for i in version]
-        recall = [i[1] for i in version]
-        f1score = [i[2] for i in version]
+    descr_y = 0.72
 
-        descr_y = 0.72
+    # Convert setting labels to numerical x positions
+    x = np.array([0.8, 1.2, 1.8, 2.2, 3])
+    x_manual = np.array([0.8, 1.8])
+    x_automatic = np.array([1.2, 2.2, 3])
+    offset = 0.08  # horizontal shift for scatter separation
 
-        # Convert setting labels to numerical x positions
-        x = np.array([0.8, 1.2, 1.8, 2.2, 3])
-        offset = 0.08  # horizontal shift for scatter separation
+    # Plot
+    fig, ax = plt.subplots(figsize=(8, 5))
 
-        # Plot
-        fig, ax = plt.subplots(figsize=(8, 5))
+    main_label_size = 22
+    sub_label_size = 16
+    main_tick_size = 16
+    legendsize = 18
 
-        main_label_size = 22
-        sub_label_size = 16
-        main_tick_size = 16
-        legendsize = 18
+    color_pm = "#3AA67E"
+    color_pa = "#17E69A"
+    color_rm = "#438CA7"
+    color_ra = "#17AEE6"
+    color_fm = "#694BA6"
+    color_fa = "#6322E6"
 
-        plt.scatter(x - offset, precision, label="Precision", marker="o", s=80)
-        plt.scatter(x,         recall, label="Recall", marker="^", s=80)
-        plt.scatter(x + offset, f1score, label="F1-score", marker="*", s=80)
+    plt.scatter(x_manual - offset, precision_manual, label="Precision manual", color=color_pm, marker="o", s=80)
+    plt.scatter(x_manual,         recall_manual, label="Recall manual", color=color_rm, marker="o", s=80)
+    plt.scatter(x_manual + offset, f1score_manual, label="F1-score manual", color=color_fm, marker="o", s=80)
 
-        plt.text(1, descr_y, "SGN", fontsize=main_label_size, horizontalalignment="center")
-        plt.text(2, descr_y, "IHC", fontsize=main_label_size, horizontalalignment="center")
-        plt.text(3, descr_y, "Synapse", fontsize=main_label_size, horizontalalignment="center")
+    plt.scatter(x_automatic - offset, precision_automatic, label="Precision automatic", color=color_pa, marker="s", s=80)
+    plt.scatter(x_automatic,         recall_automatic, label="Recall automatic", color=color_ra, marker="s", s=80)
+    plt.scatter(x_automatic + offset, f1score_automatic, label="F1-score automatic", color=color_fa, marker="s", s=80)
 
-        # Labels and formatting
-        plt.xticks(x, setting, fontsize=sub_label_size)
-        plt.yticks(fontsize=main_tick_size)
-        plt.ylabel("Value", fontsize=main_label_size)
-        plt.ylim(0.76, 1)
-        plt.legend(loc="lower right",
-                   fontsize=legendsize)
-        plt.grid(axis="y", linestyle="--", alpha=0.5)
+    # Labels and formatting
+    plt.xticks([1,2,3], setting, fontsize=main_label_size)
+    plt.yticks(fontsize=main_tick_size)
+    plt.ylabel("Value", fontsize=main_label_size)
+    plt.ylim(0.76, 1)
+    plt.legend(loc="lower right",
+                fontsize=legendsize)
+    plt.grid(axis="y", linestyle="--", alpha=0.5)
 
-        plt.tight_layout()
-        prism_cleanup_axes(ax)
+    plt.tight_layout()
+    prism_cleanup_axes(ax)
 
-        if ".png" in save_path:
-            plt.savefig(save_path, bbox_inches="tight", pad_inches=0.1, dpi=png_dpi)
-        else:
-            plt.savefig(save_path, bbox_inches='tight', pad_inches=0)
+    if ".png" in save_path:
+        plt.savefig(save_path, bbox_inches="tight", pad_inches=0.1, dpi=png_dpi)
+    else:
+        plt.savefig(save_path, bbox_inches='tight', pad_inches=0)
 
-        if plot:
-            plt.show()
-        else:
-            plt.close()
+    if plot:
+        plt.show()
+    else:
+        plt.close()
 
 
 # Load the synapse counts for all IHCs from the relevant tables.
@@ -423,8 +415,8 @@ def main():
     os.makedirs(args.figure_dir, exist_ok=True)
 
     # Panes A and B: Qualitative comparison of visualization results.
-    fig_02a_sgn(save_dir=args.figure_dir, plot=args.plot)
-    fig_02b_ihc(save_dir=args.figure_dir, plot=args.plot)
+    # fig_02a_sgn(save_dir=args.figure_dir, plot=args.plot)
+    # fig_02b_ihc(save_dir=args.figure_dir, plot=args.plot)
 
     # Panel C: Evaluation of the segmentation results:
     fig_02c(save_path=os.path.join(args.figure_dir, f"fig_02c.{FILE_EXTENSION}"), plot=args.plot, all_versions=False)

scripts/figures/plot_fig3.py

@@ -13,11 +13,9 @@
 from matplotlib import cm, colors
 
 from flamingo_tools.s3_utils import BUCKET_NAME, create_s3_target
-from util import sliding_runlength_sum, frequency_mapping, prism_style, prism_cleanup_axes, SYNAPSE_DIR_ROOT
+from util import sliding_runlength_sum, frequency_mapping, SYNAPSE_DIR_ROOT
 
-# INPUT_ROOT = "/home/pape/Work/my_projects/flamingo-tools/scripts/M_LR_000227_R/scale3"
-INPUT_ROOT = "/mnt/vast-nhr/projects/nim00007/data/moser/cochlea-lightsheet/frequency_mapping/M_LR_000227_R/scale3"
-FILE_EXTENSION = "png"
+INPUT_ROOT = "/home/martin/Documents/lightsheet-cochlea/M_LR_000227_R"
 
 TYPE_TO_CHANNEL = {
     "Type-Ia": "CR",
@@ -93,7 +91,7 @@ def get_tonotopic_data():
         return pickle.load(f)
 
 
-def _plot_colormap(vol, title, plot, save_path):
+def _plot_colormap(vol, title, plot, save_path, cmap="viridis"):
     # before creating the figure:
     matplotlib.rcParams.update({
         "font.size": 14,          # base font size
@@ -110,10 +108,16 @@ def _plot_colormap(vol, title, plot, save_path):
 
     freq_min = np.min(np.nonzero(vol))
     freq_max = vol.max()
-    norm = colors.Normalize(vmin=freq_min, vmax=freq_max, clip=True)
-    cmap = plt.get_cmap("viridis")
+    # norm = colors.Normalize(vmin=freq_min, vmax=freq_max, clip=True)
+    norm = colors.LogNorm(vmin=freq_min, vmax=freq_max, clip=True)
+    tick_values = np.array([10, 20, 40, 80])
+
+    cmap = plt.get_cmap(cmap)
 
-    cb = plt.colorbar(cm.ScalarMappable(norm=norm, cmap=cmap), cax=ax, orientation="horizontal")
+    cb = plt.colorbar(cm.ScalarMappable(norm=norm, cmap=cmap), cax=ax, orientation="horizontal",
+                      ticks=tick_values)
+    cb.ax.xaxis.set_major_formatter(matplotlib.ticker.ScalarFormatter())
+    cb.ax.xaxis.set_minor_locator(matplotlib.ticker.NullLocator())
     cb.set_label("Frequency [kHz]")
     plt.title(title)
     plt.tight_layout()
@@ -127,19 +131,29 @@ def _plot_colormap(vol, title, plot, save_path):
     plt.close()
 
 
-def fig_03a(save_path, plot, plot_napari):
+def fig_03a(save_path, plot, plot_napari, cmap="viridis"):
     path_ihc = os.path.join(INPUT_ROOT, "frequencies_IHC_v4c.tif")
     path_sgn = os.path.join(INPUT_ROOT, "frequencies_SGN_v2.tif")
     sgn = imageio.imread(path_sgn)
     ihc = imageio.imread(path_ihc)
-    _plot_colormap(sgn, title="Tonotopic Mapping", plot=plot, save_path=save_path)
+    _plot_colormap(sgn, title="Tonotopic Mapping", plot=plot, save_path=save_path, cmap=cmap)
 
     # Show the image in napari for rendering.
     if plot_napari:
         import napari
+        from napari.utils import Colormap
+        # cmap = plt.get_cmap(cmap)
+        mpl_cmap = plt.get_cmap(cmap)
+
+        # Sample it into an array of RGBA values
+        colors = mpl_cmap(np.linspace(0, 1, 256))
+
+        # Wrap into napari Colormap
+        napari_cmap = Colormap(colors, name=f"{cmap}_custom")
+
         v = napari.Viewer()
-        v.add_image(ihc, colormap="viridis")
-        v.add_image(sgn, colormap="viridis")
+        v.add_image(ihc, colormap=napari_cmap)
+        v.add_image(sgn, colormap=napari_cmap)
         napari.run()
 
 
@@ -180,8 +194,7 @@ def fig_03c_rl(save_path, plot=False):
         plt.close()
 
 
-def fig_03c_octave(tonotopic_data, save_path, plot=False, use_alias=True, trendlines=False):
-    prism_style()
+def fig_03c_octave(tonotopic_data, save_path, plot=False, use_alias=True):
     ihc_version = "ihc_counts_v4c"
     tables = glob(os.path.join(SYNAPSE_DIR_ROOT, ihc_version, "ihc_count_M_LR*.tsv"))
     assert len(tables) == 4, len(tables)
@@ -207,55 +220,16 @@ def fig_03c_octave(tonotopic_data, save_path, plot=False, use_alias=True, trendl
     result["x_pos"] = result["octave_band"].map(band_to_x)
 
     fig, ax = plt.subplots(figsize=(8, 4))
-    trend_dict = {}
     for name, grp in result.groupby("cochlea"):
         ax.scatter(grp["x_pos"], grp["value"], label=name, s=60, alpha=0.8)
 
-        if trendlines:
-            x_positions = grp["x_pos"]
-            sorted_idx = np.argsort(x_positions)
-            x_sorted = np.array(x_positions)[sorted_idx]
-            y_sorted = np.array(grp["value"])[sorted_idx]
-            trend_dict[name] = {"x_sorted": x_sorted,
-                                "y_sorted": y_sorted,
-                                }
-
-    if trendlines:
-        def get_trendline_values(trend_dict):
-            x_sorted = [trend_dict[k]["x_sorted"] for k in trend_dict.keys()][0]
-            y_sorted_all = [trend_dict[k]["y_sorted"] for k in trend_dict.keys()]
-            y_sorted = []
-            for num in range(len(x_sorted)):
-                y_sorted.append(np.mean([y[num] for y in y_sorted_all]))
-            return x_sorted, y_sorted
-
-        # Trendline left
-        x_sorted, y_sorted = get_trendline_values(trend_dict)
-
-        trend, = ax.plot(
-            x_sorted,
-            y_sorted,
-            linestyle="dotted",
-            color="grey",
-            alpha=0.7
-        )
-
-        # trendline_legend = ax.legend(handles=[trend], loc='lower center')
-        # trendline_legend = ax.legend(
-        #     handles=[trend],
-        #     labels=["Trendline"],
-        #     loc="upper left"
-        # )
-        # # Add the legend manually to the Axes.
-        # ax.add_artist(trendline_legend)
-
     ax.set_xticks(range(len(bin_labels)))
     ax.set_xticklabels(bin_labels)
     ax.set_xlabel("Octave band (kHz)")
 
-    ax.set_ylabel("Average Ribbon Synapse Count per IHC", fontsize=10)
+    ax.set_ylabel("Average Ribbon Synapse Count per IHC")
+    ax.set_title("Ribbon synapse count per octave band")
     plt.legend(title="Cochlea")
-    prism_cleanup_axes(ax)
 
     if ".png" in save_path:
         plt.savefig(save_path, bbox_inches="tight", pad_inches=0.1, dpi=png_dpi)
@@ -345,15 +319,16 @@ def main():
     tonotopic_data = get_tonotopic_data()
 
     # Panel A: Tonotopic mapping of SGNs and IHCs (rendering in napari + heatmap)
-    # fig_03a(save_path=os.path.join(args.figure_dir, f"fig_03a_cmap.{FILE_EXTENSION}"),
-    # plot=args.plot, plot_napari=True)
+    cmap = "plasma"
+    fig_03a(save_path=os.path.join(args.figure_dir, f"fig_03a_cmap_{cmap}.{FILE_EXTENSION}"),
+            plot=args.plot, plot_napari=True, cmap=cmap)
 
     # Panel C: Spatial distribution of synapses across the cochlea.
     # We have two options: running sum over the runlength or per octave band
     # fig_03c_rl(save_path=os.path.join(args.figure_dir, f"fig_03c_runlength.{FILE_EXTENSION}"), plot=args.plot)
     fig_03c_octave(tonotopic_data=tonotopic_data,
-                   save_path=os.path.join(args.figure_dir, f"fig_03c_octave.{FILE_EXTENSION}"),
-                   plot=args.plot, trendlines=True)
+                    save_path=os.path.join(args.figure_dir, f"fig_03c_octave.{FILE_EXTENSION}"),
+                    plot=args.plot)
 
     # Panel D: Spatial distribution of SGN sub-types.
     # fig_03d_fraction(save_path=os.path.join(args.figure_dir, f"fig_03d_fraction.{FILE_EXTENSION}"), plot=args.plot)

scripts/figures/plot_fig4.py

@@ -309,16 +309,21 @@ def fig_04e(chreef_data, save_path, plot, intensity=False, gerbil=False, use_ali
     # Assign a color to each cochlea (ignoring side)
     cochleas = sorted({name_lr[:-1] for name_lr in result["cochlea"].unique()})
     colors = plt.cm.tab10.colors  # pick a colormap
+    colors = ["#DB5748", "#DB4B6F", "#DB49C6", "#B748DB", "#8748DB"]
+
     color_map = {cochlea: colors[i % len(colors)] for i, cochlea in enumerate(cochleas)}
+
+
     if len(cochleas) == 1:
         color_map = {"L": colors[0], "R": colors[1]}
 
     # Track which cochlea names we have already added to the legend
     legend_added = set()
 
+    offset = 0.02
     trend_dict = {}
 
-    for name_lr, grp in result.groupby("cochlea"):
+    for num, (name_lr, grp) in enumerate(result.groupby("cochlea")):
         name, side = name_lr[:-1], name_lr[-1]
         if len(cochleas) == 1:
             label_name = name_lr
@@ -327,7 +332,7 @@ def fig_04e(chreef_data, save_path, plot, intensity=False, gerbil=False, use_ali
             label_name = name
             color = color_map[name]
 
-        x_positions = grp["x_pos"] + offset_map[side]
+        x_positions = grp["x_pos"] + offset_map[side] - len(cochleas) / 2 * offset + offset * num
         ax.scatter(
             x_positions,
             grp["value"],

scripts/figures/util.py

@@ -31,6 +31,20 @@ def ax_prism_boxplot(ax, data, positions=None, color="tab:blue"):
     return bp
 
 
+prism_palette = [
+    "#4E79A7",  # blue
+    "#F28E2B",  # orange
+    "#E15759",  # red
+    "#76B7B2",  # teal
+    "#59A14F",  # green
+    "#EDC948",  # yellow
+    "#B07AA1",  # purple
+    "#FF9DA7",  # pink
+    "#9C755F",  # brown
+    "#BAB0AC"   # gray
+]
+
+
 def prism_style():
     plt.style.use("default")  # reset any active styles
     plt.rcParams.update({
@@ -44,6 +58,7 @@ def prism_style():
         "axes.linewidth": 1.2,
         "axes.labelsize": 14,
         "axes.labelweight": "bold",
+        "axes.prop_cycle": plt.cycler("color", prism_palette),
 
         # Ticks
         "xtick.direction": "out",