support exonNumber and strand in gencode genes

ssadedin · Jul 15, 2024 · b0d3fde · b0d3fde
1 parent 6c398ee
commit b0d3fde
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 7 deletions.
diff --git a/src/main/groovy/gngs/gencode/Gencode.groovy b/src/main/groovy/gngs/gencode/Gencode.groovy
@@ -120,7 +120,7 @@ class Gencode implements GeneAnnotationSource {
         final String regionId = attributes['ID']
         if(type == 'gene') {
             region.properties.remove('attributes')
-            Gene gene = new Gene(region, regionId, (String)attributes.gene_name)
+            Gene gene = new Gene(region, regionId, (String)attributes.gene_name, region['strand'] as char)
             if(attributes.hgnc_id)
                 gene.hgnc_id = ((String)attributes.hgnc_id).split(':')[-1]
             gene.type = (String)attributes.gene_type
@@ -133,7 +133,8 @@ class Gencode implements GeneAnnotationSource {
         // coding exons appear both with the exon and CDS designation
         // UNLESS they are part of the UTR
         if(type == 'CDS' || type == 'three_prime_UTR' || type == 'five_prime_UTR') {
-            feature = new Exon(region, regionId)
+            int exonNumber = Integer.parseInt((String)attributes.exon_number)
+            feature = new Exon(region, regionId, exonNumber)
             if(type == 'CDS')
                 feature.coding = true
         }

diff --git a/src/main/groovy/gngs/gencode/Gene.groovy b/src/main/groovy/gngs/gencode/Gene.groovy
@@ -47,11 +47,13 @@ class Gene extends Feature<Transcript> {
     String symbol
     String hgnc_id
     String type
+    char strand
 
 //    Transcript transcript
-    public Gene(IRegion region, String id, String symbol) {
+    public Gene(IRegion region, String id, String symbol, char strand) {
         super(region, id, null);
         this.symbol = symbol
+        this.strand = strand
     }
 }
 
@@ -67,7 +69,8 @@ class Exon extends Feature {
     int exonNumber
     Gene gene
     boolean coding
-    public Exon(IRegion region, String id) {
+    public Exon(IRegion region, String id, int exonNumber) {
         super(region, id, null);
+        this.exonNumber = exonNumber
     }
 }
diff --git a/src/test/groovy/gngs/GencodeTest.groovy b/src/test/groovy/gngs/GencodeTest.groovy
@@ -43,11 +43,17 @@ class GencodeTest {
         println "Exons of the main transcript: "
 
         dvl1.children[0].children.each {
-            println "Exon $it"
+            println "Exon $it ($it.exonNumber, coding=$it.coding)"
         }
 
+        assert dvl1.strand == '-'
         assert dvl1.children[0].children.size() == 17
         assert dvl1.children[1].children.size() == 17
+
+        // DVL a case where coding sequence ends within the exon,
+        // hence exon 15 represented twice, once with coding sequence,
+        // once without
+        assert dvl1.children[0].children[0].exonNumber == 15
     }
 
     @Test
@@ -70,6 +76,7 @@ class GencodeTest {
 //            dvl_regions.save("dvl_regions_${i}.bed", extra: { tx.id })
 //        }
 
+        assert dvl1.strand == '-'
         assert dvl1.children[0].children.size() == 17
         assert dvl1.children[1].children.size() == 17
     }
@@ -101,9 +108,9 @@ class GencodeTest {
 
         println "CDS for DVL1 is $cds"
 
-        def expected = [TAS1R3:0, DVL1:2103, MIR6808:0]
+        def expected = [TAS1R3:0, DVL1:2073, MIR6808:0]
         assert cds.size() == 3
-        assert cds.DVL1 == 2073
+        assert cds.DVL1 == expected.DVL1
         assert cds.TAS1R3 == 0
         assert cds.MIR6808 == 0