Refactor FC histogram plotting func

Author: phuongho43

ezplot/plot_cytometry.py

@@ -6,11 +6,10 @@
 import pandas as pd
 import seaborn as sns
 
-from ezplot.process import combine_rcg, load_fcs
 from ezplot.style import PALETTE, STYLE
 
 
-def plot_fc_hist(fig_fp, y_csv_fp, gateline, class_labels, group_labels, xlabel, figsize=(32, 16), palette=PALETTE, rc_params=STYLE):
+def plot_fc_hist(fig_fp, y_df, gateline, class_labels, group_labels, xlabel, figsize=(32, 16), palette=PALETTE, rc_params=STYLE):
     fig_fp = Path(fig_fp)
     fig_fp.parent.mkdir(parents=True, exist_ok=True)
     with sns.axes_style("whitegrid"), mpl.rc_context(rc_params):
@@ -28,55 +27,52 @@ def plot_fc_hist(fig_fp, y_csv_fp, gateline, class_labels, group_labels, xlabel,
                 "legend.fontsize": 36,
             },
         )
-        y_df = pd.read_csv(Path(y_csv_fp))
         y_df = y_df.loc[(y_df["response"] > 0)]
         fg = sns.FacetGrid(y_df, palette=palette, row="group", hue="repeat", aspect=7, height=2)
         fg.map_dataframe(sns.kdeplot, x="response", hue="class", fill=True, alpha=0.25, log_scale=True, palette=palette, linewidth=0)
         fg.map_dataframe(sns.kdeplot, x="response", hue="class", fill=False, alpha=1, log_scale=True, palette=palette, linewidth=1)
         fg.refline(y=0, linewidth=1, linestyle="-", color="#212121", clip_on=False)
         fg.refline(x=gateline, color="#212121", clip_on=False)
-        for g, ax in enumerate(fg.axes.ravel()):
-            yg_c0_df = y_df.loc[(y_df["group"] == g) & (y_df["class"] == 0)]
-            yg_c1_df = y_df.loc[(y_df["group"] == g) & (y_df["class"] == 1)]
-            tot_c0 = len(yg_c0_df)
-            tot_c1 = len(yg_c1_df)
-            pos_c0 = np.sum(yg_c0_df["response"] > gateline)
-            pos_c1 = np.sum(yg_c1_df["response"] > gateline)
-            pct_c0 = np.around(100 * (pos_c0 / tot_c0), 2)
-            pct_c1 = np.around(100 * (pos_c1 / tot_c1), 2)
-            ax.text(1, 0.30, f"{pct_c0}%", color=palette[0], fontsize=36, ha="right", transform=ax.transAxes)
-            ax.text(1, 0.02, f"{pct_c1}%", color=palette[1], fontsize=36, ha="right", transform=ax.transAxes)
-            ax.text(0, 0.02, f"{group_labels[g]}", color="#212121", fontsize=36, ha="left", transform=ax.transAxes)
-            if g >= 2:
-                ax.text(0, 0.30, "Decoder +", color="#212121", fontsize=36, ha="left", transform=ax.transAxes)
-        handles = [
-            mpl.lines.Line2D([], [], color=palette[0], lw=6),
-            mpl.lines.Line2D([], [], color=palette[1], lw=6),
-        ]
-        fg.axes.ravel()[0].legend(handles, class_labels, loc=(0, 0.4))
+        handles = []
+        for c in range(y_df["class"].nunique()):
+            for g, ax in enumerate(fg.axes.ravel()):
+                ycg_df = y_df.loc[(y_df["group"] == g) & (y_df["class"] == c)]
+                tot_ycg = len(ycg_df)
+                pos_ycg = np.sum(ycg_df["response"] > gateline)
+                pct_ycg = np.around(100 * (pos_ycg / tot_ycg), 2)
+                ax.text(1, c * 0.30 + 0.02, f"{pct_ycg}%", color=palette[c], fontsize=36, ha="right", transform=ax.transAxes)
+                ax.text(0, 0.02, f"{group_labels[g]}", color="#212121", fontsize=36, ha="left", transform=ax.transAxes)
+            handles.append(mpl.lines.Line2D([], [], color=palette[c], lw=6))
+        fg.axes.ravel()[0].legend(handles, class_labels, loc=(0, 0.62))
         fg.fig.subplots_adjust(hspace=0)
         fg.set_titles("")
-        fg.set(yticks=[], ylabel="", xlim=(1, 1e7))
+        fg.set(yticks=[], ylabel="", xlim=(1, None))
         fg.set_xlabels(xlabel, fontsize=48, fontweight="bold")
         fg.despine(left=True, bottom=True)
         fg.fig.savefig(fig_fp)
         plt.close("all")
 
 
 def main():
-    save_csv_fp = "/home/phuong/data/phd-project/3--antigen/0--K562-fc-staining/y.csv"
-    data_dp = "/home/phuong/data/phd-project/3--antigen/0--K562-fc-staining/data/"
-    combine_rcg(data_dp, load_fnc=load_fcs, load_fnc_kwargs={"channel": "FL4_A"}, save_csv_fp=save_csv_fp)
+    # save_csv_fp = "/home/phuong/data/phd-project/3--antigen/0--K562-fc-staining/y.csv"
+    # data_dp = "/home/phuong/data/phd-project/3--antigen/0--K562-fc-staining/data/"
+    # combine_rcg(data_dp, load_fnc=load_fcs, load_fnc_kwargs={"channel": "FL4_A"}, save_csv_fp=save_csv_fp)
 
-    # fig_fp = "/home/phuong/data/phd-project/3--antigen/0--K562-fc-staining/y.png"
-    # y_csv_fp = "/home/phuong/data/phd-project/3--antigen/0--K562-fc-staining/y.csv"
-    # group_labels = [r"$\mathdefault{Plain\ K562}$", r"$\mathdefault{Constitutive}$", r"$\mathdefault{None\ Input}$", r"$\mathdefault{Sparse\ Input}$", r"$\mathdefault{Dense\ Input}$"]
-    # class_labels = [r"$\alpha$-CD19 AF647", r"$\alpha$-PSMA APC"]
-    # palette = ["#8069EC", "#EA822C"]
-    # xlabel = "Fluorescence (AU)"
-    # fig_fp = "/home/phuong/data/phd-project/figures/fig_4b.png"
-    # gateline = 3e3
-    # plot_fc_hist(fig_fp, y_csv_fp, gateline, class_labels, group_labels, xlabel, figsize=(32, 16), palette=palette)
+    fig_fp = "/home/phuong/data/phd-project/figures/fig_4b.png"
+    y_csv_fp = "/home/phuong/data/phd-project/3--antigen/0--K562-fc-staining/y.csv"
+    y_df = pd.read_csv(y_csv_fp)
+    group_labels = [
+        "Plain K562",
+        "Constitutive",
+        "Decoder + \nNone Input",
+        "Decoder + \nSparse Input",
+        "Decoder + \nDense Input",
+    ]
+    class_labels = [r"$\alpha$-CD19 AF647", r"$\alpha$-PSMA APC"]
+    palette = ["#8069EC", "#EA822C"]
+    xlabel = "Fluorescence (AU)"
+    gateline = 3e3
+    plot_fc_hist(fig_fp, y_df, gateline, class_labels, group_labels, xlabel, figsize=(32, 16), palette=palette)
 
 
 if __name__ == "__main__":