🎨🔧 improve swarmplots, add methods in ALD comp.

RasmussenLab · Nov 26, 2023 · c9e00e4 · c9e00e4
1 parent 29a549a
commit c9e00e4
Show file tree

Hide file tree

Showing 4 changed files with 42 additions and 148 deletions.
diff --git a/project/10_4_ald_compare_single_pg.ipynb b/project/10_4_ald_compare_single_pg.ipynb
@@ -20,6 +20,7 @@
    "source": [
     "from pathlib import Path\n",
     "\n",
+    "import logging\n",
     "import matplotlib.pyplot as plt\n",
     "import pandas as pd\n",
     "\n",
@@ -32,10 +33,10 @@
     "import vaep.imputation\n",
     "\n",
     "logger = vaep.logging.setup_nb_logger()\n",
-    "\n",
+    "logging.getLogger('fontTools').setLevel(logging.WARNING)\n",
     "\n",
     "plt.rcParams['figure.figsize'] = [4, 2.5]  # [16.0, 7.0] , [4, 3]\n",
-    "vaep.plotting.make_large_descriptors(5)"
+    "vaep.plotting.make_large_descriptors(7)"
    ]
   },
   {
@@ -580,7 +581,9 @@
    "cell_type": "code",
    "execution_count": null,
    "id": "f813f693",
-   "metadata": {},
+   "metadata": {
+    "lines_to_next_cell": 2
+   },
    "outputs": [],
    "source": [
     "min_y_int, max_y_int = vaep.plotting.data.get_min_max_iterable(\n",
@@ -592,66 +595,6 @@
     "min_max, target_name"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "d03628eb",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "for idx in feat_sel:\n",
-    "    fig, ax = plt.subplots()\n",
-    "\n",
-    "    feat_observed = data[idx].dropna()\n",
-    "\n",
-    "    label_template = '{method} (N={n:,d}, q={q:.3f})'\n",
-    "    # observed data\n",
-    "    vaep.plotting.data.plot_histogram_intensities(\n",
-    "        feat_observed,\n",
-    "        ax=ax,\n",
-    "        min_max=min_max,\n",
-    "        label=label_template.format(method='measured',\n",
-    "                                    n=len(feat_observed),\n",
-    "                                    q=float(qvalues.loc[idx, ('None', 'qvalue')])),\n",
-    "        color='grey',\n",
-    "        alpha=0.6)\n",
-    "\n",
-    "    # all models\n",
-    "    for i, method in enumerate(model_keys):\n",
-    "        try:\n",
-    "            pred = pred_real_na.loc[pd.IndexSlice[:, idx], method].dropna()\n",
-    "            if len(pred) == 0:\n",
-    "                # in case no values was imputed -> qvalue is as based on measured\n",
-    "                label = label_template.format(method=method,\n",
-    "                                              n=len(pred),\n",
-    "                                              q=float(qvalues.loc[idx, ('None', 'qvalue')]\n",
-    "                                                      ))\n",
-    "            else:\n",
-    "                label = label_template.format(method=method,\n",
-    "                                              n=len(pred),\n",
-    "                                              q=float(qvalues.loc[idx, (method, 'qvalue')]\n",
-    "                                                      ))\n",
-    "            ax, bins = vaep.plotting.data.plot_histogram_intensities(\n",
-    "                pred,\n",
-    "                ax=ax,\n",
-    "                min_max=min_max,\n",
-    "                label=label,\n",
-    "                color=f'C{i}',\n",
-    "                alpha=0.6)\n",
-    "        except KeyError:\n",
-    "            print(f\"No missing values for {idx}: {method}\")\n",
-    "            continue\n",
-    "    first_pg = idx.split(\";\")[0]\n",
-    "    ax.set_title(\n",
-    "        f'Imputation for protein group {first_pg} with target {target_name} (N= {len(data):,d} samples)')\n",
-    "    ax.set_ylabel('count measurments')\n",
-    "    _ = ax.legend()\n",
-    "    files_out[fname.name] = fname.as_posix()\n",
-    "    vaep.savefig(\n",
-    "        fig, folder / f'{first_pg}_hist.pdf')\n",
-    "    plt.close(fig)"
-   ]
-  },
   {
    "cell_type": "markdown",
    "id": "b9db8a0e",
@@ -682,8 +625,8 @@
     "    tmp_dot.remove()\n",
     "\n",
     "    feat_observed = data[idx].dropna()\n",
-    "    label_template = '{method} (N={n:,d}, q={q:.3f})'\n",
-    "    key = label_template.format(method='measured',\n",
+    "    label_template = '{method}\\n(N={n:,d}, q={q:.3f})'\n",
+    "    key = label_template.format(method='observed',\n",
     "                                n=len(feat_observed),\n",
     "                                q=float(qvalues.loc[idx, ('None', 'qvalue')])\n",
     "                                )\n",
@@ -726,7 +669,7 @@
     "                           order=groups_order,\n",
     "                           dodge=True,\n",
     "                           hue=args.target,\n",
-    "                           size=1,\n",
+    "                           size=2,\n",
     "                           ax=ax)\n",
     "    first_pg = idx.split(\";\")[0]\n",
     "    ax.set_title(\n",
@@ -747,7 +690,6 @@
     "    _ = ax.collections[0].set_paths([new_mk])\n",
     "    _ = ax.collections[1].set_paths([new_mk])\n",
     "\n",
-    "    # import matplotlib.lines as mlines\n",
     "    label_target_0, label_target_1 = ax.collections[-2].get_label(), ax.collections[-1].get_label()\n",
     "    _ = ax.collections[-2].set_label(f'imputed, {label_target_0}')\n",
     "    _ = ax.collections[-1].set_label(f'imputed, {label_target_1}')\n",

diff --git a/project/10_4_ald_compare_single_pg.py b/project/10_4_ald_compare_single_pg.py
@@ -21,6 +21,7 @@
 # %%
 from pathlib import Path
 
+import logging
 import matplotlib.pyplot as plt
 import pandas as pd
 
@@ -33,10 +34,10 @@
 import vaep.imputation
 
 logger = vaep.logging.setup_nb_logger()
-
+logging.getLogger('fontTools').setLevel(logging.WARNING)
 
 plt.rcParams['figure.figsize'] = [4, 2.5]  # [16.0, 7.0] , [4, 3]
-vaep.plotting.make_large_descriptors(5)
+vaep.plotting.make_large_descriptors(7)
 
 # %% [markdown]
 # ## Parameters
@@ -313,58 +314,6 @@
 
 min_max, target_name
 
-# %%
-for idx in feat_sel:
-    fig, ax = plt.subplots()
-
-    feat_observed = data[idx].dropna()
-
-    label_template = '{method} (N={n:,d}, q={q:.3f})'
-    # observed data
-    vaep.plotting.data.plot_histogram_intensities(
-        feat_observed,
-        ax=ax,
-        min_max=min_max,
-        label=label_template.format(method='measured',
-                                    n=len(feat_observed),
-                                    q=float(qvalues.loc[idx, ('None', 'qvalue')])),
-        color='grey',
-        alpha=0.6)
-
-    # all models
-    for i, method in enumerate(model_keys):
-        try:
-            pred = pred_real_na.loc[pd.IndexSlice[:, idx], method].dropna()
-            if len(pred) == 0:
-                # in case no values was imputed -> qvalue is as based on measured
-                label = label_template.format(method=method,
-                                              n=len(pred),
-                                              q=float(qvalues.loc[idx, ('None', 'qvalue')]
-                                                      ))
-            else:
-                label = label_template.format(method=method,
-                                              n=len(pred),
-                                              q=float(qvalues.loc[idx, (method, 'qvalue')]
-                                                      ))
-            ax, bins = vaep.plotting.data.plot_histogram_intensities(
-                pred,
-                ax=ax,
-                min_max=min_max,
-                label=label,
-                color=f'C{i}',
-                alpha=0.6)
-        except KeyError:
-            print(f"No missing values for {idx}: {method}")
-            continue
-    first_pg = idx.split(";")[0]
-    ax.set_title(
-        f'Imputation for protein group {first_pg} with target {target_name} (N= {len(data):,d} samples)')
-    ax.set_ylabel('count measurments')
-    _ = ax.legend()
-    files_out[fname.name] = fname.as_posix()
-    vaep.savefig(
-        fig, folder / f'{first_pg}_hist.pdf')
-    plt.close(fig)
 
 # %% [markdown]
 # ## Compare with target annotation
@@ -383,8 +332,8 @@
     tmp_dot.remove()
 
     feat_observed = data[idx].dropna()
-    label_template = '{method} (N={n:,d}, q={q:.3f})'
-    key = label_template.format(method='measured',
+    label_template = '{method}\n(N={n:,d}, q={q:.3f})'
+    key = label_template.format(method='observed',
                                 n=len(feat_observed),
                                 q=float(qvalues.loc[idx, ('None', 'qvalue')])
                                 )
@@ -427,7 +376,7 @@
                            order=groups_order,
                            dodge=True,
                            hue=args.target,
-                           size=1,
+                           size=2,
                            ax=ax)
     first_pg = idx.split(";")[0]
     ax.set_title(
@@ -448,7 +397,6 @@
     _ = ax.collections[0].set_paths([new_mk])
     _ = ax.collections[1].set_paths([new_mk])
 
-    # import matplotlib.lines as mlines
     label_target_0, label_target_1 = ax.collections[-2].get_label(), ax.collections[-1].get_label()
     _ = ax.collections[-2].set_label(f'imputed, {label_target_0}')
     _ = ax.collections[-1].set_label(f'imputed, {label_target_1}')

diff --git a/project/config/appl_ald_data/plasma/proteinGroups/comparison.yaml b/project/config/appl_ald_data/plasma/proteinGroups/comparison.yaml
@@ -1,8 +1,8 @@
 folder_experiment: runs/appl_ald_data/plasma/proteinGroups
 out_folder: diff_analysis # subfolder of experiment folder
-fn_rawfile_metadata: 'data/ALD_study/processed/raw_meta.csv'
+fn_rawfile_metadata: "data/ALD_study/processed/raw_meta.csv"
 make_plots: True
-covar: 
+covar:
   kleiner: age,bmi,gender_num,nas_steatosis_ordinal,abstinent_num
   inflammation: age,bmi,gender_num,nas_steatosis_ordinal,abstinent_num
   steatosis: age,bmi,gender_num,abstinent_num,kleiner,nas_inflam
@@ -19,9 +19,10 @@ annotaitons_gene_col: PG.Genes
 baseline: RSN
 ref_method_score:
 methods:
- - Median
- - CF
- - DAE
- - VAE
- - rf
- - KNN
+  - Median
+  - CF
+  - DAE
+  - VAE
+  - QRILC
+  - TRKNN
+  - RF
diff --git a/project/config/appl_ald_data/plasma/proteinGroups/config.yaml b/project/config/appl_ald_data/plasma/proteinGroups/config.yaml
@@ -12,24 +12,27 @@ models:
   - VAE
   - KNN
 NAGuideR_methods:
-  - ZERO
-  - MINIMUM
+  - BPCA
   - COLMEDIAN
-  - ROWMEDIAN
+  - GSIMP
+  - IMPSEQ
+  - IMPSEQROB
+  - IRM
   - KNN_IMPUTE
-  # - SEQKNN # error
-  - BPCA
-  - SVDMETHOD
   - LLS
-  - MLE
-  - QRILC
+  # - MICE-CART # stopped after 24h
+  # - MICE-NORM # stopped after 24h
   - MINDET
+  - MINIMUM
   - MINPROB
-  - IRM
-  # - IMPSEQ # error
-  - IMPSEQROB
-  # - MICE-NORM # stopped after 30mins
-  # - MICE-CART # stopped after 30mins
-  # - TRKNN # error
+  - MLE
+  - MSIMPUTE
+  - MSIMPUTE_MNAR
+  - PI
+  - QRILC
   - RF
-  - PI
+  - ROWMEDIAN
+  # - SEQKNN # Error in x[od, ismiss, drop = FALSE]: subscript out of bounds
+  - SVDMETHOD
+  - TRKNN
+  - ZERO