track hub per tissue

Author: tjiagoM

14_track_hub.ipynb

@@ -106,33 +106,42 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [],
    "source": [
     "with open('track_hub/hg38/trackDb.txt', 'w') as f_track_hubs:\n",
     "    for tissue in TISSUES:\n",
+    "        all_tissue_genes = []\n",
     "        try:\n",
     "            for community_id in range(1, 999999):\n",
     "                arr_com = []\n",
     "                dic_community = pickle.load(open(\"svm_results/\" + tissue + '_' + str(community_id) + \".pkl\", \"rb\"))\n",
-    "                len_common = len(dic_community['genes'])\n",
-    "\n",
-    "                with open(f'track_hub/hg38/{tissue}_{community_id}.bed', 'w') as f:\n",
-    "                    for gene in dic_community['genes']:\n",
-    "                        if gene in dic_all_genes_info.keys():\n",
-    "                            gene_info = dic_all_genes_info[gene]\n",
-    "                            f.write(f'{gene_info[\"chr\"]}\\t{gene_info[\"chr_start\"]}\\t{gene_info[\"chr_end\"]}\\n')\n",
-    "                \n",
-    "                f_track_hubs.write(f'track {tissue}_{community_id}\\n')\n",
-    "                f_track_hubs.write(f'bigDataUrl https://raw.githubusercontent.com/tjiagoM/gtex-transcriptome-modelling/master/track_hub/hg38/{tissue}_{community_id}.bb\\n')\n",
-    "                f_track_hubs.write(f'shortLabel {tissue}_{community_id}\\n')\n",
-    "                f_track_hubs.write(f'longLabel {tissue}_{community_id}\\n')\n",
-    "                f_track_hubs.write(f'type bigBed\\n')\n",
-    "                f_track_hubs.write(f'\\n')\n",
-    "\n",
+    "                all_tissue_genes.extend(dic_community['genes'])\n",
     "        except Exception as e:\n",
-    "            pass"
+    "            pass\n",
+    "        \n",
+    "        # Removing duplicates\n",
+    "        all_tissue_genes = list(set(all_tissue_genes))\n",
+    "        \n",
+    "        # Saving all the tissue's genes\n",
+    "        with open(f'track_hub/hg38/{tissue}.bed', 'w') as f:\n",
+    "            for gene in all_tissue_genes:\n",
+    "                if gene in dic_all_genes_info.keys():\n",
+    "                    gene_info = dic_all_genes_info[gene]\n",
+    "                    f.write(f'{gene_info[\"chr\"]}\\t{gene_info[\"chr_start\"]}\\t{gene_info[\"chr_end\"]}\\n')\n",
+    "                # Some genes are separated with a '-' rather than with a '.'\n",
+    "                elif gene.replace('.', '-') in dic_all_genes_info.keys():\n",
+    "                    gene_info = dic_all_genes_info[gene.replace('.', '-')]\n",
+    "                    f.write(f'{gene_info[\"chr\"]}\\t{gene_info[\"chr_start\"]}\\t{gene_info[\"chr_end\"]}\\n')\n",
+    "\n",
+    "        f_track_hubs.write(f'track {tissue}\\n')\n",
+    "        f_track_hubs.write(f'bigDataUrl https://raw.githubusercontent.com/tjiagoM/gtex-transcriptome-modelling/master/track_hub/hg38/{tissue}.bb\\n')\n",
+    "        f_track_hubs.write(f'shortLabel {tissue}\\n')\n",
+    "        f_track_hubs.write(f'longLabel {tissue}\\n')\n",
+    "        f_track_hubs.write(f'type bigBed\\n')\n",
+    "        f_track_hubs.write(f'visibility full\\n')\n",
+    "        f_track_hubs.write(f'\\n')"
    ]
   },
   {