update

Author: Bill Majoros

Gene.py

@@ -30,6 +30,7 @@
 #   strand=gene.getStrand()
 #   substrate=gene.getSubstrate()
 #   gff=gene.toGff()
+#   exons=gene.getMergedExons()
 #   
 ######################################################################
 
@@ -38,6 +39,27 @@ def __init__(self):
         self.transcripts=[]
         self.transcriptHash={}
 
+    def getMergedExons(self):
+        transcripts=self.transcripts
+        exons=[]
+        for transcript in transcripts:
+            raw=transcript.getRawExons()
+            exons.extend(raw)
+            #print("RAW:",len(raw))
+            #for i in range(len(raw)):
+                #print("\t",raw[i].begin,raw[i].end)
+            #print()
+        exons.sort(key=lambda x: x.begin)
+        n=len(exons)
+        i=0
+        while(i<n-1):
+            if(exons[i].overlaps(exons[i+1])):
+                exons[i].end=max(exons[i].end,exons[i+1].end)
+                del exons[i+1]
+                n-=1
+            else: i+=1
+        return exons
+
     def getStrand(self):
         transcripts=self.transcripts
         transcript=transcripts[0]

test-transcript-reader.py

@@ -15,9 +15,21 @@
 #filename="/home/bmajoros/1000G/assembly/local-genes.gff"
 #filename="/home/bmajoros/1000G/assembly/tmp.gff"
 #filename="test/data/tmp.gff"
-filename="test/data/local-genes.gff"
+#filename="test/data/local-genes.gff"
+filename="/home/bmajoros/ensembl/protein-coding.gff"
 
 reader=GffTranscriptReader()
+genes=reader.loadGenes(filename)
+for gene in genes:
+    exons=gene.getMergedExons()
+    unmerged=0
+    for transcript in gene.transcripts:
+        unmerged+=len(transcript.getRawExons())
+    print(unmerged,"exons merged to",len(exons))
+    #for i in range(len(exons)):
+    #    print("MERGED TO:",exons[i].begin,exons[i].end)
+    #    print()
+
 #transcripts=reader.loadGFF(filename)
 #for transcript in transcripts:
     #print(transcript.getID())
@@ -49,8 +61,8 @@
 #for key in keys:
 #    print(key)
 
-hashTable=reader.loadGeneIdHash(filename)
-keys=hashTable.keys()
-for key in keys:
-    print(key)
+#hashTable=reader.loadGeneIdHash(filename)
+#keys=hashTable.keys()
+#for key in keys:
+#    print(key)