FEAT #83

jass/models/inittable.py

@@ -375,6 +375,7 @@ def add_gene_annotation(
     gene_start = []
     gene_end = []
     gene_direction = []
+    gene_biotype = []
 
     # lists containing exon data
     exon_id_label = []
@@ -414,6 +415,23 @@ def add_gene_annotation(
                     decode_id_4 = decode_id_3[0].split("=")
                     decode_id_5 = decode_id_4[1].split(":")
                     gene_GeneID.append(decode_id_5[1])
+                    decode_id_6 = decode_id_1[5].split("=")
+                    if decode_id_6[0] == "gene_biotype":
+                        The_biotype = decode_id_6[1].rstrip('\r\n')
+                        gene_biotype.append(The_biotype)
+                    else:
+                        decode_id_7 = decode_id_1[6].split("=")
+                        if decode_id_7[0] == "gene_biotype":
+                            The_biotype = decode_id_7[1].rstrip('\r\n')
+                            gene_biotype.append(The_biotype)
+                        else:
+                            decode_id_8 = decode_id_1[7].split("=")
+                            if decode_id_8[0] == "gene_biotype":
+                                The_biotype = decode_id_8[1].rstrip('\r\n')
+                                gene_biotype.append(The_biotype)
+                            else:
+                                The_biotype = "unknown"
+                                gene_biotype.append(The_biotype)
 
                 elif elements[2] == "exon":
                     TMP__exon_chr.append(chr)
@@ -447,6 +465,7 @@ def add_gene_annotation(
             "start": gene_start,
             "end": gene_end,
             "direction": gene_direction,
+            "biotype": gene_biotype,
         }
     )
     df_exon = DataFrame(