add smoke status comparision

04PreneoplasiaEvolution/01PathomicsTrends/explore_pathomics_trend_smoke.py

@@ -0,0 +1,46 @@
+# -*- coding: utf-8 -*-
+
+import os, sys
+import argparse, shutil, json
+import numpy as np
+import pandas as pd
+import matplotlib.pyplot as plt
+import seaborn as sns
+
+
+def set_args():
+    parser = argparse.ArgumentParser(description = "Extract lesion cellular ratio  features")
+    parser.add_argument("--data_root",        type=str,       default="/Data")
+    parser.add_argument("--combine_dir",      type=str,       default="CombineAnalysis")
+    parser.add_argument("--trend_dir",        type=str,       default="PathomicsTrend")
+    parser.add_argument("--plot_format",      type=str,       default=".png", choices=[".png", ".pdf"])      
+    parser.add_argument("--rand_seed",        type=int,       default=1234)    
+
+    args = parser.parse_args()
+    return args
+
+
+if __name__ == "__main__":
+    args = set_args()
+    np.random.seed(args.rand_seed)
+
+    combine_dir = os.path.join(args.data_root, args.combine_dir)
+    path_trend_dir = os.path.join(combine_dir, args.trend_dir, "Smoke")
+    if not os.path.exists(path_trend_dir):
+        os.makedirs(path_trend_dir)
+
+    fuse_fea_path = os.path.join(combine_dir, "CombineROIFeatures.csv")
+    fuse_fea_df = pd.read_csv(fuse_fea_path)
+
+    # compare two cohort features
+    pathomics_lst = ["AEC-Ratio", "LYM-Ratio", "AEC-Density", "LYM-Density", "Altieri2-Entropy", "Altieri3-Entropy",
+        "AEC-Contrast",	"AEC-Energy", "LYM-Contrast", "LYM-Energy"]
+    stage_list = ["Normal", "AAH", "AIS", "MIA", "ADC"]    
+    for path_fea in pathomics_lst:
+        fig, axes = plt.subplots(1, 1, figsize=(8, 5))
+        sns.violinplot(data=fuse_fea_df, x = "Stages", y=path_fea, hue = "SmokeStatus", order=stage_list,  orient='v')
+        # axes.set_ylim(0, 1.0)
+        plt.legend(loc = "best")
+        plt.tight_layout()
+        trend_plot_path = os.path.join(path_trend_dir, "{}-Trend-Smoke{}".format(path_fea, args.plot_format))
+        plt.savefig(trend_plot_path, transparent=False, dpi=300)