make logo plots narrower and number lineplot sites better (#34)

dms-vep · Nov 14, 2024 · 399b313 · 399b313
1 parent c6807a8
commit 399b313
Show file tree

Hide file tree

Showing 14 changed files with 91,513 additions and 91,919 deletions.
diff --git a/docs/notebooks/escape_logos.html b/docs/notebooks/escape_logos.html
diff --git a/notebooks/escape_logos.py.ipynb b/notebooks/escape_logos.py.ipynb
@@ -94,12 +94,20 @@
     "# amino acids to plot\n",
     "aas = list(dmslogo.colorschemes.AA_CHARGE)\n",
     "\n",
+    "# line plot xtick locations\n",
+    "lineplot_xticks = [0, 100, 200, 300, 400]\n",
+    "\n",
     "# make logo plots with each of these antibody groups in addition to per-antibody logos\n",
     "antibody_groups = {\n",
     "    \"region_3_antibodies\": [\"17C7\", \"CR4098\", \"RVA122\", \"RVC58\"],\n",
     "    \"region_1_antibodies\": [\"RVC20\", \"CR57\"],   \n",
     "}\n",
     "\n",
+    "# sites to exclude from highlighting, do not show these sites even if they\n",
+    "# meet highlight criteria. Key by antibody with value lists of sites\n",
+    "sites_no_highlight = {\n",
+    "}\n",
+    "\n",
     "# file extensions on saved plots\n",
     "file_extensions = [\".svg\", \".pdf\"]"
    ]
@@ -209,6 +217,17 @@
     "            ).astype(int)\n",
     "        )\n",
     "    )\n",
+    "    .merge(\n",
+    "        (\n",
+    "            pd.DataFrame(sites_no_highlight.items(), columns=[\"antibody\", \"site\"])\n",
+    "            .explode(\"site\")\n",
+    "            .assign(manual_no_highlight=True)\n",
+    "        ),\n",
+    "        on=[\"antibody\", \"site\"],\n",
+    "        validate=\"many_to_one\",\n",
+    "        how=\"left\",\n",
+    "    )\n",
+    "    .assign(manual_no_highlight=lambda x: x[\"manual_no_highlight\"].fillna(False))\n",
     ")"
    ]
   },
@@ -220,7 +239,7 @@
    "outputs": [],
    "source": [
     "# indicate the sites to highlight\n",
-    "per_site_escape[\"highlight\"] = functools.reduce(\n",
+    "per_site_escape[\"highlight\"] = ~per_site_escape[\"manual_no_highlight\"] & functools.reduce(\n",
     "    operator.or_,\n",
     "    [\n",
     "        (\n",
@@ -264,7 +283,7 @@
     "            header=alt.Header(orient=\"bottom\", labelFontSize=11, labelFontStyle=\"bold\"),\n",
     "        ),\n",
     "        alt.Color(\"highlight\", title=\"site to highlight\"),\n",
-    "        tooltip=[\"site\", \"rank\", \"antibody\"],\n",
+    "        tooltip=[\"site\", \"rank\", \"antibody\", \"manual_no_highlight\"],\n",
     "        strokeWidth=alt.condition(site_selection, alt.value(3), alt.value(0.25)),\n",
     "    )\n",
     "    .mark_circle(size=40, strokeOpacity=1, fillOpacity=0.4, stroke=\"black\")\n",
@@ -402,7 +421,7 @@
     "        .query(\"antibody in @antibodies_to_plot\")\n",
     "        .assign(highlight=lambda x: x.groupby(\"site\")[\"highlight\"].transform(\"any\"))\n",
     "    )\n",
-    "    fig, _ = dmslogo.facet_plot(\n",
+    "    fig, axes = dmslogo.facet_plot(\n",
     "        df,\n",
     "        x_col=\"sequential_site\",\n",
     "        show_col=\"highlight\",\n",
@@ -412,22 +431,30 @@
     "            \"height_col2\": None if positive_escape_only else \"negative_site_escape\",\n",
     "            \"xtick_col\": \"site\",\n",
     "            \"ylabel\": \"escape\",\n",
-    "            \"widthscale\": 0.4,\n",
+    "            \"widthscale\": 0.3,\n",
     "        },\n",
     "        draw_logo_kwargs={\n",
     "            \"letter_col\": \"mutant\",\n",
     "            \"letter_height_col\": \"escape\",\n",
     "            \"xtick_col\": \"site\",\n",
     "            \"color_col\": \"letter_color\",\n",
-    "            \"widthscale\": 0.75,\n",
+    "            \"widthscale\": 0.7,\n",
     "        },\n",
     "        share_xlabel=True,\n",
     "        share_ylabel=True,\n",
     "        height_per_ax=1.9,\n",
-    "        wspace=0.7,\n",
+    "        wspace=0.6,\n",
     "        share_ylim_across_rows=False,\n",
     "    )\n",
     "\n",
+    "    # adjust x-ticks on line plots\n",
+    "    assert axes.shape[1] == 2\n",
+    "    for i in range(axes.shape[0]):\n",
+    "        axes[i, 0].xaxis.set_ticks(lineplot_xticks)\n",
+    "        if i == axes.shape[0] - 1:\n",
+    "            axes[i, 0].xaxis.set_ticklabels(lineplot_xticks)\n",
+    "        \n",
+    "\n",
     "    for ext in file_extensions:\n",
     "        fname = os.path.join(logoplot_subdir, f\"{antibody_group}{ext}\")\n",
     "        print(f\"Saving to {fname}\")\n",
@@ -436,14 +463,6 @@
     "    display(fig)\n",
     "    plt.close(fig)"
    ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "d3a497c8-23fd-448d-ae9d-5eafd9963ec5",
-   "metadata": {},
-   "outputs": [],
-   "source": []
   }
  ],
  "metadata": {