Fix TN/TP/FN/FP plots

Clinical-Genomics-Lund · Jul 26, 2024 · bee6395 · bee6395
1 parent d954014
commit bee6395
Showing 1 changed file with 8 additions and 10 deletions.
diff --git a/notebooks/tb_validation.ipynb b/notebooks/tb_validation.ipynb
@@ -365,7 +365,7 @@
    "outputs": [],
    "source": [
     "# Initialize a DataFrame to hold the results\n",
-    "results = pd.DataFrame(columns=['True positives', 'True negatives', 'False positives', 'False negatives', 'Not tested', 'Unknown'])\n",
+    "results = pd.DataFrame(columns=['True positives', 'True negatives', 'False positives', 'False negatives'])\n",
     "\n",
     "ass_loa = []\n",
     "n_counts = []\n",
@@ -376,7 +376,7 @@
     "    false_negatives = ((km_tb_df[col1] == \"R\") & (km_tb_df[col2].isna())).sum()\n",
     "    not_tested = (km_tb_df[col1].isin([\"ej testad\", \"Ej testad\", np.nan])).sum()\n",
     "    unknown = (~km_tb_df[col1].isin([\"ej testad\", \"Ej testad\", \"R\", \"S\", np.nan])).sum()\n",
-    "    total_count = len(km_tb_df)\n",
+    "    total_count = true_positives + true_negatives + false_positives + false_negatives\n",
     "    n_counts.append(total_count)\n",
     "    if true_positives + true_negatives + false_positives + false_negatives == 0:\n",
     "        accuracy = 0.0\n",
@@ -397,25 +397,23 @@
     "        \"specificity\": specificity\n",
     "    })\n",
     "    results.loc[f'{col1} vs {col2}'] = [\n",
-    "        true_positives / total_count * 100,\n",
-    "        true_negatives / total_count * 100,\n",
-    "        false_positives / total_count * 100,\n",
-    "        false_negatives / total_count * 100,\n",
-    "        not_tested / total_count * 100,\n",
-    "        unknown / total_count * 100\n",
+    "        true_positives,\n",
+    "        true_negatives,\n",
+    "        false_positives,\n",
+    "        false_negatives\n",
     "    ]\n",
     "\n",
     "# Write out csv\n",
     "results.to_csv(tb_validation_accuracy_output, index=True)\n",
     "\n",
     "# Plotting\n",
     "fig, ax = plt.subplots(figsize=(10, 8))\n",
-    "results.plot(kind='barh', stacked=True, ax=ax, color=['green', 'gold', 'orange', 'red', 'darkblue', 'lightblue'])\n",
+    "results.plot(kind='barh', stacked=True, ax=ax, color=['green', 'gold', 'orange', 'red'])\n",
     "\n",
     "# Customizing plot\n",
     "ax.set_xlabel('Percentage (%)')\n",
     "ax.set_title('AMR calling accuracy')\n",
-    "ax.legend(['True positives', 'True negatives', 'False positives', 'False negatives', 'Not tested', 'Unknown'], bbox_to_anchor=(1.05, 1), loc='upper left')\n",
+    "ax.legend(['True positives', 'True negatives', 'False positives', 'False negatives'], bbox_to_anchor=(1.05, 1), loc='upper left')\n",
     "ax.set_xlim(0, 110)\n",
     "for i in range(len(results)):\n",
     "    ax.text(100, i, f'n={n_counts[i]}', ha='left', va='center', fontsize=10, color='black')\n",