Merge pull request #42 from jenna-tomkinson/add_main_figure_3

Add main figure 3

.gitattributes

@@ -2,3 +2,5 @@
 1.train_models/classify_genotypes/data/nf1_model_pre_evaluation_results.parquet filter=lfs diff=lfs merge=lfs -text
 2.evaluate_models/log_reg_plates_3_3p_5_cp_norm_data/model_eval_data/plate_precision_recall.parquet filter=lfs diff=lfs merge=lfs -text
 2.evaluate_models/classify_genotypes/model_evaluation_data/precision_recall.parquet filter=lfs diff=lfs merge=lfs -text
+1.train_models/classify_genotypes/model_data.parquet filter=lfs diff=lfs merge=lfs -text
+1.train_models/classify_genotypes/data/nf1_eval_data.parquet filter=lfs diff=lfs merge=lfs -text

1.train_models/classify_genotypes/classify_genotypes.ipynb

@@ -115,16 +115,31 @@
    },
    "outputs": [],
    "source": [
-    "plate5df_path = pathlib.Path(root_dir / \"nf1_painting_repo/3.processing_features/data/single_cell_profiles/Plate_5_sc_feature_selected.parquet\").resolve(strict=True)\n",
-    "plate3df_path = pathlib.Path(root_dir / \"nf1_painting_repo/3.processing_features/data/single_cell_profiles/Plate_3_sc_feature_selected.parquet\").resolve(strict=True)\n",
-    "plate3pdf_path = pathlib.Path(root_dir / \"nf1_painting_repo/3.processing_features/data/single_cell_profiles/Plate_3_prime_sc_feature_selected.parquet\").resolve(strict=True)\n",
-    "plate4df_path = pathlib.Path(root_dir / \"nf1_painting_repo/3.processing_features/data/single_cell_profiles/Plate_4_sc_feature_selected.parquet\").resolve(strict=True)\n",
+    "plate5df_path = pathlib.Path(\n",
+    "    root_dir\n",
+    "    / \"../nf1_cellpainting_data/3.processing_features/data/single_cell_profiles/Plate_5_sc_feature_selected.parquet\"\n",
+    ").resolve(strict=True)\n",
+    "plate3df_path = pathlib.Path(\n",
+    "    root_dir\n",
+    "    / \"../nf1_cellpainting_data/3.processing_features/data/single_cell_profiles/Plate_3_sc_feature_selected.parquet\"\n",
+    ").resolve(strict=True)\n",
+    "plate3pdf_path = pathlib.Path(\n",
+    "    root_dir\n",
+    "    / \"../nf1_cellpainting_data/3.processing_features/data/single_cell_profiles/Plate_3_prime_sc_feature_selected.parquet\"\n",
+    ").resolve(strict=True)\n",
+    "plate4df_path = pathlib.Path(\n",
+    "    root_dir\n",
+    "    / \"../nf1_cellpainting_data/3.processing_features/data/single_cell_profiles/Plate_4_sc_feature_selected.parquet\"\n",
+    ").resolve(strict=True)\n",
     "\n",
     "plate5df = pd.read_parquet(plate5df_path)\n",
     "plate4df = pd.read_parquet(plate4df_path)\n",
     "plate3df = pd.read_parquet(plate3df_path)\n",
     "plate3pdf = pd.read_parquet(plate3pdf_path)\n",
     "\n",
+    "# Correct Plate_3_prime plate column bug\n",
+    "plate3pdf[\"Metadata_Plate\"] = \"Plate_3_prime\"\n",
+    "\n",
     "# Set the seed\n",
     "rng = np.random.default_rng(0)"
    ]
@@ -368,11 +383,44 @@
    "outputs": [],
    "source": [
     "# Columns common to all plates\n",
-    "plate_cols = list(set(plate5df.columns) & set(plate3df.columns) & set(plate3pdf.columns) & set(plate4df.columns))\n",
+    "plate_cols = list(\n",
+    "    set(plate5df.columns)\n",
+    "    & set(plate3df.columns)\n",
+    "    & set(plate3pdf.columns)\n",
+    "    & set(plate4df.columns)\n",
+    ")\n",
+    "\n",
+    "restdf = pd.concat(\n",
+    "    [\n",
+    "        rest5df[plate_cols],\n",
+    "        rest3df[plate_cols],\n",
+    "        rest3pdf[plate_cols],\n",
+    "        rest4df[plate_cols],\n",
+    "    ],\n",
+    "    ignore_index=True,\n",
+    ").reset_index(drop=True)\n",
+    "\n",
+    "# Add Metadata_datasplit column to restdf\n",
+    "restdf = restdf.assign(datasplit='rest')\n",
+    "\n",
+    "testdf = pd.concat(\n",
+    "    [\n",
+    "        test5df[plate_cols],\n",
+    "        test3df[plate_cols],\n",
+    "        test3pdf[plate_cols],\n",
+    "        test4df[plate_cols],\n",
+    "    ],\n",
+    "    ignore_index=True,\n",
+    ").reset_index(drop=True)\n",
+    "\n",
+    "# Add Metadata_datasplit column to testdf\n",
+    "testdf = testdf.assign(datasplit='test')\n",
     "\n",
-    "restdf = pd.concat([rest5df[plate_cols], rest3df[plate_cols], rest3pdf[plate_cols], rest4df[plate_cols]], ignore_index=True).reset_index(drop=True)\n",
+    "# Concatenate restdf and testdf vertically\n",
+    "model_df = pd.concat([restdf, testdf], ignore_index=True)\n",
     "\n",
-    "testdf = pd.concat([test5df[plate_cols], test3df[plate_cols], test3pdf[plate_cols], test4df[plate_cols]], ignore_index=True).reset_index(drop=True)"
+    "# Save model data to apply the model for evaluation\n",
+    "model_df.to_parquet(\"./model_data.parquet\")"
    ]
   },
   {