Refactor plotting code (#137)

* move remaining analysis logic from plot.py to analyze.py * add `Model.model_label` * move repeated plotting code to new `PlotData` * doc `ModelHash`->`Model` in analysis.rst * convert logo text to paths * change font size spec for axis vis; rename `NORMAL_NODE_COLOR` -> `DEFAULT_NODE_COLOT` --------- Co-authored-by: Daniel Weindl <[email protected]>
PEtab-dev · Jan 7, 2025 · 85c355c · 85c355c
1 parent 6c5d621
commit 85c355c
Show file tree

Hide file tree

Showing 16 changed files with 898 additions and 559 deletions.
diff --git a/doc/analysis.rst b/doc/analysis.rst
@@ -9,3 +9,20 @@ to get a quick overview over all models, as a pandas dataframe.
 
 Additionally, see the Python API docs for the :mod:`petab_select.analyze` module, which contains some methods to subset and group models,
 or compute "weights" (e.g. Akaike weights).
+
+Model hashes
+^^^^^^^^^^^^
+
+Model hashes are special objects in the library, that are generated from model-specific information that is unique within a single PEtab Select problem.
+
+This means you can reconstruct the model given some model hash. For example, with this model hash `M1-000`, you can reconstruct the :class:`petab_select.ModelHash` from a string, then reconstruct the :class:`petab_select.Model`.
+
+.. code-block:: language
+
+   ModelHash.from_hash("M1-000").get_model(petab_select_problem)
+
+You can use this to get the uncalibrated version of a calibrated model.
+
+.. code-block:: language
+
+   model.hash.get_model(petab_select_problem)
diff --git a/doc/examples/calibrated_models/calibrated_models.yaml b/doc/examples/calibrated_models/calibrated_models.yaml
@@ -4,16 +4,16 @@
   estimated_parameters:
     sigma_x2: 4.462298422134608
   iteration: 1
-  model_hash: M_0-000
-  model_id: M_0-000
-  model_subspace_id: M_0
+  model_hash: M-000
+  model_id: M-000
+  model_subspace_id: M
   model_subspace_indices:
   - 0
   - 0
   - 0
   parameters:
-    k1: 0
-    k2: 0
+    k1: 0.2
+    k2: 0.1
     k3: 0
   model_subspace_petab_yaml: petab_problem.yaml
   predecessor_model_hash: virtual_initial_model-
@@ -24,59 +24,59 @@
     k3: 0.0
     sigma_x2: 0.12242920113658338
   iteration: 2
-  model_hash: M_1-000
-  model_id: M_1-000
-  model_subspace_id: M_1
+  model_hash: M-001
+  model_id: M-001
+  model_subspace_id: M
   model_subspace_indices:
   - 0
   - 0
-  - 0
+  - 1
   parameters:
     k1: 0.2
     k2: 0.1
     k3: estimate
   model_subspace_petab_yaml: petab_problem.yaml
-  predecessor_model_hash: M_0-000
+  predecessor_model_hash: M-000
 - criteria:
     AICc: -0.27451438069575573
     NLLH: -4.137257190347878
   estimated_parameters:
     k2: 0.10147824307890803
     sigma_x2: 0.12142219599557078
   iteration: 2
-  model_hash: M_2-000
-  model_id: M_2-000
-  model_subspace_id: M_2
+  model_hash: M-010
+  model_id: M-010
+  model_subspace_id: M
   model_subspace_indices:
   - 0
-  - 0
+  - 1
   - 0
   parameters:
     k1: 0.2
     k2: estimate
     k3: 0
   model_subspace_petab_yaml: petab_problem.yaml
-  predecessor_model_hash: M_0-000
+  predecessor_model_hash: M-000
 - criteria:
     AICc: -0.7053270766271886
     NLLH: -4.352663538313594
   estimated_parameters:
     k1: 0.20160925279667963
     sigma_x2: 0.11714017664827497
   iteration: 2
-  model_hash: M_3-000
-  model_id: M_3-000
-  model_subspace_id: M_3
+  model_hash: M-100
+  model_id: M-100
+  model_subspace_id: M
   model_subspace_indices:
-  - 0
+  - 1
   - 0
   - 0
   parameters:
     k1: estimate
     k2: 0.1
     k3: 0
   model_subspace_petab_yaml: petab_problem.yaml
-  predecessor_model_hash: M_0-000
+  predecessor_model_hash: M-000
 - criteria:
     AICc: 9.294672923372811
     NLLH: -4.352663538313594
@@ -85,19 +85,19 @@
     k3: 0.0
     sigma_x2: 0.11714017664827497
   iteration: 3
-  model_hash: M_5-000
-  model_id: M_5-000
-  model_subspace_id: M_5
+  model_hash: M-101
+  model_id: M-101
+  model_subspace_id: M
   model_subspace_indices:
+  - 1
   - 0
-  - 0
-  - 0
+  - 1
   parameters:
     k1: estimate
     k2: 0.1
     k3: estimate
   model_subspace_petab_yaml: petab_problem.yaml
-  predecessor_model_hash: M_3-000
+  predecessor_model_hash: M-100
 - criteria:
     AICc: 7.8521704398854
     NLLH: -5.0739147800573
@@ -106,19 +106,19 @@
     k2: 0.0859052351446815
     sigma_x2: 0.10386846319370771
   iteration: 3
-  model_hash: M_6-000
-  model_id: M_6-000
-  model_subspace_id: M_6
+  model_hash: M-110
+  model_id: M-110
+  model_subspace_id: M
   model_subspace_indices:
-  - 0
-  - 0
+  - 1
+  - 1
   - 0
   parameters:
     k1: estimate
     k2: estimate
     k3: 0
   model_subspace_petab_yaml: petab_problem.yaml
-  predecessor_model_hash: M_3-000
+  predecessor_model_hash: M-100
 - criteria:
     AICc: 35.94352968170024
     NLLH: -6.028235159149878
@@ -128,16 +128,16 @@
     k3: 0.0010850434974038557
     sigma_x2: 0.08859278245811462
   iteration: 4
-  model_hash: M_7-000
-  model_id: M_7-000
-  model_subspace_id: M_7
+  model_hash: M-111
+  model_id: M-111
+  model_subspace_id: M
   model_subspace_indices:
-  - 0
-  - 0
-  - 0
+  - 1
+  - 1
+  - 1
   parameters:
     k1: estimate
     k2: estimate
     k3: estimate
   model_subspace_petab_yaml: petab_problem.yaml
-  predecessor_model_hash: M_3-000
+  predecessor_model_hash: M-100
diff --git a/doc/examples/calibrated_models/model_space.tsv b/doc/examples/calibrated_models/model_space.tsv
@@ -1,9 +1,2 @@
 model_subspace_id	petab_yaml	k1	k2	k3
-M_0	petab_problem.yaml	0	0	0
-M_1	petab_problem.yaml	0.2	0.1	estimate
-M_2	petab_problem.yaml	0.2	estimate	0
-M_3	petab_problem.yaml	estimate	0.1	0
-M_4	petab_problem.yaml	0.2	estimate	estimate
-M_5	petab_problem.yaml	estimate	0.1	estimate
-M_6	petab_problem.yaml	estimate	estimate	0
-M_7	petab_problem.yaml	estimate	estimate	estimate
+M	petab_problem.yaml	0.2;estimate	0.1;estimate	0
diff --git a/doc/examples/visualization.ipynb b/doc/examples/visualization.ipynb
@@ -31,7 +31,6 @@
     "\n",
     "import petab_select\n",
     "import petab_select.plot\n",
-    "from petab_select import VIRTUAL_INITIAL_MODEL\n",
     "\n",
     "models = petab_select.Models.from_yaml(\n",
     "    \"calibrated_models/calibrated_models.yaml\"\n",
@@ -45,7 +44,9 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "models.df.style.background_gradient(\n",
+    "models.df.drop(\n",
+    "    columns=[petab_select.Criterion.AIC, petab_select.Criterion.BIC]\n",
+    ").style.background_gradient(\n",
     "    cmap=matplotlib.colormaps.get_cmap(\"summer\"),\n",
     "    subset=[petab_select.Criterion.AICC],\n",
     ")"
@@ -56,7 +57,7 @@
    "id": "aaeb0606",
    "metadata": {},
    "source": [
-    "We customize the labels here to just be the second part of a model hash: a binary string that describes their estimated parameters. We additionally set a custom color for a couple of models, and the default color for the other models. We also set a nicer label for the \"virtual initial model\", which is a hypothetical model that PEtab Select uses by default to initialize a forward search (with no parameters)."
+    "To use the plotting methods, we need to first setup an object that contains information common to multiple plotting methods. This can include the models, custom colors and labels, and the criterion."
    ]
   },
   {
@@ -66,25 +67,21 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# Custom labels\n",
-    "labels = {}\n",
-    "for model in models:\n",
-    "    labels[model.hash] = \"M_\" + \"\".join(\n",
-    "        \"1\" if value == petab_select.ESTIMATE else \"0\"\n",
-    "        for value in model.parameters.values()\n",
-    "    )\n",
-    "labels[VIRTUAL_INITIAL_MODEL.hash] = \"\\n\".join(\n",
-    "    petab_select.VIRTUAL_INITIAL_MODEL.model_subspace_id.split(\"_\")\n",
-    ").title()\n",
-    "\n",
     "# Custom colors for some models\n",
     "colors = {\n",
-    "    \"M_000\": \"lightgreen\",\n",
-    "    \"M_001\": \"lightgreen\",\n",
+    "    \"M-000\": \"lightgreen\",\n",
+    "    \"M-001\": \"lightgreen\",\n",
     "}\n",
     "\n",
+    "plot_data = petab_select.plot.PlotData(\n",
+    "    models=models,\n",
+    "    criterion=petab_select.Criterion.AICC,\n",
+    "    colors=colors,\n",
+    "    relative_criterion=True,\n",
+    ")\n",
+    "\n",
     "# Change default color\n",
-    "petab_select.plot.NORMAL_NODE_COLOR = \"darkgray\""
+    "petab_select.plot.DEFAULT_NODE_COLOR = \"darkgray\""
    ]
   },
   {
@@ -106,7 +103,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "petab_select.plot.upset(models=models, criterion=petab_select.Criterion.AICC);"
+    "petab_select.plot.upset(plot_data=plot_data);"
    ]
   },
   {
@@ -128,11 +125,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "petab_select.plot.line_best_by_iteration(\n",
-    "    models=models,\n",
-    "    criterion=petab_select.Criterion.AICC,\n",
-    "    labels=labels,\n",
-    ");"
+    "petab_select.plot.line_best_by_iteration(plot_data=plot_data);"
    ]
   },
   {
@@ -152,12 +145,11 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "petab_select.plot.graph_history(\n",
-    "    models=models,\n",
-    "    criterion=petab_select.Criterion.AICC,\n",
-    "    labels=labels,\n",
-    "    colors=colors,\n",
-    ");"
+    "# Add the relative criterion value to each label\n",
+    "plot_data.augment_labels(criterion=True)\n",
+    "petab_select.plot.graph_history(plot_data=plot_data)\n",
+    "# Reset the labels (remove the relative criterion)\n",
+    "plot_data.augment_labels()"
    ]
   },
   {
@@ -177,12 +169,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "petab_select.plot.bar_criterion_vs_models(\n",
-    "    models=models,\n",
-    "    criterion=petab_select.Criterion.AICC,\n",
-    "    labels=labels,\n",
-    "    colors=colors,\n",
-    ");"
+    "petab_select.plot.bar_criterion_vs_models(plot_data=plot_data);"
    ]
   },
   {
@@ -207,10 +194,7 @@
    "outputs": [],
    "source": [
     "petab_select.plot.scatter_criterion_vs_n_estimated(\n",
-    "    models=models,\n",
-    "    criterion=petab_select.Criterion.AICC,\n",
-    "    labels=labels,\n",
-    "    colors=colors,\n",
+    "    plot_data=plot_data,\n",
     "    # Uncomment to turn off jitter.\n",
     "    # max_jitter=0,\n",
     ");"
@@ -231,7 +215,7 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "5ce191fc",
+   "id": "21157e4d-b2ba-4cb1-95f6-e14052c86959",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -244,10 +228,11 @@
     "# cmap = matplotlib.colors.LinearSegmentedColormap.from_list(\"\", [\"green\",\"lightgreen\"])\n",
     "# colorbar_mappable = matplotlib.cm.ScalarMappable(norm=norm, cmap=cmap)\n",
     "\n",
+    "# Augment labels with the changes in parameters of each model, compared to their predecessor model\n",
+    "plot_data.augment_labels(added_parameters=True, removed_parameters=True)\n",
+    "\n",
     "petab_select.plot.graph_iteration_layers(\n",
-    "    models=models,\n",
-    "    criterion=petab_select.Criterion.AICC,\n",
-    "    labels=labels,\n",
+    "    plot_data=plot_data,\n",
     "    draw_networkx_kwargs={\n",
     "        \"arrowstyle\": \"-|>\",\n",
     "        \"node_shape\": \"s\",\n",

diff --git a/doc/logo/editable/README.md b/doc/logo/editable/README.md
@@ -0,0 +1 @@
+The text in these copies is still text. In the copies in the parent directory, the text is converted to paths (e.g. `Path -> Object to Path` in Inkscape), for consistent rendering on systems that don't have the fonts installed.
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1 @@
		The text in these copies is still text. In the copies in the parent directory, the text is converted to paths (e.g. `Path -> Object to Path` in Inkscape), for consistent rendering on systems that don't have the fonts installed.