Updated config, updated skript for changes in Ensembl Release 95 temp…

…orary fix for #51. needs more improvement of config handling
loosolab · Jan 10, 2019 · a161ed0 · a161ed0
1 parent 13a610f
commit a161ed0
Show file tree

Hide file tree

Showing 4 changed files with 54 additions and 29 deletions.
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/ActivityCategorizer.py b/bin/3.1_create_gtf/Modules/Ensembl/ActivityCategorizer.py
@@ -120,20 +120,22 @@ def activity_comparator(self, aliaslist):
         # return_value: Array of Activitystatus by category (type in config)
 
         concatenated_array = bytearray([])
-
         length = len(self.activity[aliaslist[0]])
         input_arrays = [self.activity[index] for index in aliaslist]
-        for x in range(length):
-            if any(y[x] == 0 for y in input_arrays):
-                concatenated_array.append(0)
-            elif any(y[x] == 1 for y in input_arrays):
-                concatenated_array.append(1)
-            elif any(y[x] == 2 for y in input_arrays):
-                concatenated_array.append(2)
-            elif any(y[x] == 3 for y in input_arrays):
-                concatenated_array.append(3)
-            elif any(y[x] == 4 for y in input_arrays):
-                concatenated_array.append(4)
+        try:
+            for x in range(length):
+                if any(y[x] == 0 for y in input_arrays):
+                    concatenated_array.append(0)
+                elif any(y[x] == 1 for y in input_arrays):
+                    concatenated_array.append(1)
+                elif any(y[x] == 2 for y in input_arrays):
+                    concatenated_array.append(2)
+                elif any(y[x] == 3 for y in input_arrays):
+                    concatenated_array.append(3)
+                elif any(y[x] == 4 for y in input_arrays):
+                    concatenated_array.append(4)
+        except IndexError:
+            print("Indexerror occured")
         return concatenated_array
 
 

diff --git a/bin/3.1_create_gtf/Modules/Ensembl/GTFGen.py b/bin/3.1_create_gtf/Modules/Ensembl/GTFGen.py
@@ -89,9 +89,9 @@ def generate_additional_information(gene_id, activity):
         #                  activity = List of activity-data for specified gene
         # return_value: String for attributes (column 9) in gtf-format
 
-        if gene_id.startswith("ID=regulatory_region:"):
+        if not gene_id.startswith("ID=E"):
             gene_id = 'gene_id "'+gene_id.split(':')[1]+'"'
-        elif gene_id.startswith("ID=E"):
+        else:
             gene_id = 'gene_id "'+gene_id.split('=')[1]+'"'
 
         activity_string = 'activity "'+', '.join(activity)+'"'
@@ -107,7 +107,10 @@ def generate_activity_list(self, activity, index):
 
         activity_list = []
         for key, value in activity.items():
-            activity_list.append(key+">"+self.value_map[value[index]])
+            try:
+                activity_list.append(key+">"+self.value_map[value[index]])
+            except IndexError:
+                pass
         return activity_list
 
     def get_gtf(self, release, activity):

diff --git a/bin/3.1_create_gtf/RegGTFExtractor.py b/bin/3.1_create_gtf/RegGTFExtractor.py
diff --git a/bin/3.1_create_gtf/config/celltypes_homo_sapiens.json b/bin/3.1_create_gtf/config/celltypes_homo_sapiens.json
@@ -2,7 +2,7 @@
     {
       "type": "A549",
       "alias_ucsc": [],
-      "alias_ensembl": ["A549"]
+      "alias_ensembl": ["A549", "A673"]
     },
     {
       "type": "Aorta",
@@ -17,22 +17,32 @@
     {
       "type": "B-Cells",
       "alias_ucsc": [],
-      "alias_ensembl": ["B_cells_PB_Roadmap", "naive_B_cell_VB", "GM12878"]
+      "alias_ensembl": ["B_cells_PB_Roadmap", "naive_B_cell_VB", "GM12878", "B_cell_ENCSR682AXR", "CD38__naive_B_cell_VB", "naive_B_cell_To"]
     },
     {
       "type": "T-Cell",
       "alias_ucsc": [],
-      "alias_ensembl": ["CD4_ab_T_cell_VB", "CM_CD4_ab_T_cell_VB", "CD8_ab_T_cell_CB", "T_cells_PB_Roadmap"]
+      "alias_ensembl": ["CD4_ab_T_cell_VB",
+                        "CM_CD4_ab_T_cell_VB",
+                        "CD8_ab_T_cell_CB",
+                        "T_cells_PB_Roadmap",
+                        "CD4_ab_T_cell_CB",
+                        "CD4_positive__alpha_beta_memory_T_cell",
+                        "CD4_positive__alpha_beta_T_cell",
+                        "CD4_positive__alpha_beta_T_cell_ENCSR948ZKZ",
+                        "CD4_positive__CD25_positive__alpha_beta_regulatory_T_cell",
+                        "effector_memory_CD4_positive__alpha_beta_T_cell", "EM_CD8_ab_T_cell_VB",
+                        "naive_thymus_derived_CD4_positive__alpha_beta_T_cell", "T_helper_17_cell"]
     },
     {
       "type": "Monocyte",
       "alias_ucsc": [],
-      "alias_ensembl": ["CD14CD16__monocyte_CB", "CD14CD16__monocyte_VB", "Monocytes_CD14", "Monocytes_CD14_PB_Roadmap"]
+      "alias_ensembl": ["CD14CD16__monocyte_CB", "CD14CD16__monocyte_VB", "Monocytes_CD14", "Monocytes_CD14_PB_Roadmap", "CD14_positive_monocyte"]
     },
     {
       "type": "Neutrophil",
       "alias_ucsc": [],
-      "alias_ensembl": ["neutrophil_CB", "neutrophil_myelocyte_BM", "neutrophil_VB"]
+      "alias_ensembl": ["neutrophil_CB", "neutrophil_myelocyte_BM", "neutrophil_VB", "neutrophil"]
     },
      {
       "type": "Eosinophil",
@@ -59,7 +69,7 @@
     {
       "type": "Intestine",
       "alias_ucsc": [],
-      "alias_ensembl": ["Fetal_Intestine_Large", "Fetal_Intestine_Small", "Small_Intestine"]
+      "alias_ensembl": ["Fetal_Intestine_Large", "Fetal_Intestine_Small", "Small_Intestine", "sigmoid_colon"]
     },
     {
       "type": "AdrenalGland",
@@ -69,7 +79,7 @@
     {
       "type": "Muscle",
       "alias_ucsc": ["limb"],
-      "alias_ensembl": ["Fetal_Muscle_Leg", "Fetal_Muscle_Trunk", "Psoas_Muscle", "HSMM", "HSMMtube"]
+      "alias_ensembl": ["Fetal_Muscle_Leg", "Fetal_Muscle_Trunk", "Psoas_Muscle", "HSMM", "HSMMtube", "skeletal_muscle_myoblast"]
     },
     {
       "type": "Gastric",
@@ -79,17 +89,17 @@
     {
       "type": "Endothelial",
       "alias_ucsc": ["blood vessels"],
-      "alias_ensembl": ["EPC_VB", "HMEC", "HUVEC", "HUVEC_prol_CB", "NHEK"]
+      "alias_ensembl": ["EPC_VB", "HMEC", "HUVEC", "HUVEC_prol_CB", "NHEK", "endothelial_cell_of_umbilical_vein"]
     },
     {
       "type": "StemCells",
       "alias_ucsc": [],
-      "alias_ensembl": ["H1ESC", "H1_mesenchymal", "H1_neuronal_progenitor", "H1_trophoblast", "H9", "MSC_VB", "iPS_20b", "iPS_DF_6_9", "iPS_DF_19_11"]
+      "alias_ensembl": ["H1ESC", "HUES48", "HUES6", "HUES64",  "H1_hESC", "H1_hESC_ENCSR820QMS", "H9_ENCSR323FKB", "H1_mesenchymal", "H1_neuronal_progenitor", "H1_trophoblast", "H9", "MSC_VB", "iPS_20b", "iPS_15b", "iPS_DF_6_9", "iPS_DF_19_11", "common_myeloid_progenitor__CD34_positive", "common_myeloid_progenitor__CD34_positive_ENCSR337XXD_1", "common_myeloid_progenitor__CD34_positive_ENCSR722JRY"]
     },
     {
       "type": "Lung",
       "alias_ucsc": [],
-      "alias_ensembl": ["Lung", "IMR90", "NHLF"]
+      "alias_ensembl": ["Lung", "IMR90", "NHLF", "lung_ENCSR465WKM"]
     },
     {
       "type": "Pancreas",
@@ -99,7 +109,7 @@
     {
       "type": "Liver",
       "alias_ucsc": ["liver"],
-      "alias_ensembl": []
+      "alias_ensembl": ["hepatocyte"]
     },
     {
       "type": "Ovary",
@@ -119,7 +129,7 @@
     {
       "type": "Heart",
       "alias_ucsc": ["heart"],
-      "alias_ensembl": ["Right_Atrium", "Left_Ventricle"]
+      "alias_ensembl": ["Right_Atrium", "Left_Ventricle", "cardiac_muscle_cell", "heart_right_ventricle"]
     },
     {
       "type": "Osteoblast",
@@ -129,7 +139,7 @@
     {
       "type": "Fibroblast",
       "alias_ucsc": [],
-      "alias_ensembl": ["NHDF_AD"]
+      "alias_ensembl": ["NHDF_AD", "fibroblast_of_dermis", "fibroblast_of_lung", "IMR_90"]
     },
     {
       "type": "NK-Cells",
@@ -144,7 +154,7 @@
     {
       "type": "Brain",
       "alias_ucsc": ["midbrain (mesencephalon)", "trigeminal V (ganglion, cranial)", "forebrain", "neural tube", "hindbrain (rhombencephalon)", "dorsal root ganglion", "cranial nerve"],
-      "alias_ensembl": ["NH_A"]
+      "alias_ensembl": ["NH_A", "astrocyte", "bipolar_neuron", "brain", "neural_progenitor_cell", "neural_stem_progenitor_cell", "neuron"]
     },
     {
       "type": "Mesenchym",
@@ -175,5 +185,15 @@
       "type": "Melanocytes",
       "alias_ucsc": ["melanocytes"],
       "alias_ensembl": []
+    },
+    {
+      "type": "Miscelanious",
+      "alias_ucsc": [],
+      "alias_ensembl": ["endodermal_cell", "esophagus", "HCT116", "Karpas_422", "keratinocyte", "mammary_epithelial_cell", "MCF_7", "MM_1S", "myotube", "PC_3", "PC_9", "SK_N_SH"]
+    },
+    {
+      "type": "Kidney",
+      "alias_ucsc": [],
+      "alias_ensembl": ["kidney"]
     }
 ]