lib: add splice builder to compute splice scores based on SpliceAI, #558

opencb · Jul 12, 2021 · 617e64c · 617e64c
1 parent 28364b2
commit 617e64c
Show file tree

Hide file tree

Showing 7 changed files with 716 additions and 1 deletion.
diff --git a/cellbase-app/src/main/java/org/opencb/cellbase/app/cli/admin/AdminCliOptionsParser.java b/cellbase-app/src/main/java/org/opencb/cellbase/app/cli/admin/AdminCliOptionsParser.java
@@ -94,7 +94,7 @@ public class BuildCommandOptions {
 
         @Parameter(names = {"-d", "--data"}, description = "Comma separated list of data to build: genome, genome_info, "
                 + "gene, variation, variation_functional_score, regulation, protein, ppi, conservation, drug, "
-                + "clinical_variants, repeats, svs. 'all' builds everything.", required = true, arity = 1)
+                + "clinical_variants, repeats, svs, splice. 'all' builds everything.", required = true, arity = 1)
         public String data;
 
         @Parameter(names = {"-s", "--species"}, description = "Name of the species to be built, valid formats include 'Homo sapiens' or 'hsapiens'", required = false, arity = 1)

diff --git a/...e-app/src/main/java/org/opencb/cellbase/app/cli/admin/executors/BuildCommandExecutor.java b/...e-app/src/main/java/org/opencb/cellbase/app/cli/admin/executors/BuildCommandExecutor.java
@@ -155,6 +155,9 @@ public void execute() {
                         case EtlCommons.OBO_DATA:
                             parser = buildObo();
                             break;
+                        case EtlCommons.SPLICE_DATA:
+                            parser = buildSplice();
+                            break;
                         default:
                             logger.error("Build option '" + buildCommandOptions.data + "' is not valid");
                             break;
@@ -371,4 +374,12 @@ private Path getFastaReferenceGenome() {
         }
         return fastaFile;
     }
+
+    private CellBaseBuilder buildSplice() {
+        Path genePath = buildFolder.resolve("gene.json.gz");
+        Path genomeInfoPath = buildFolder.resolve("genome_info.json");
+        Path fastaPath = getFastaReferenceGenome();
+        CellBaseFileSerializer serializer = new CellBaseJsonFileSerializer(buildFolder, "splice");
+        return new SpliceBuilder(genePath, genomeInfoPath, fastaPath, serializer);
+    }
 }
diff --git a/cellbase-core/src/main/resources/configuration.yml b/cellbase-core/src/main/resources/configuration.yml
@@ -168,6 +168,7 @@ species:
         - regulation
         - repeats
         - variation_functional_score
+        - splice
       shards:
         - collection: "variation"
           key:

diff --git a/cellbase-lib/src/main/java/org/opencb/cellbase/lib/EtlCommons.java b/cellbase-lib/src/main/java/org/opencb/cellbase/lib/EtlCommons.java
@@ -46,6 +46,7 @@ public class EtlCommons {
     public static final String PROTEIN_DATA = "protein";
     public static final String CONSERVATION_DATA = "conservation";
     public static final String CLINICAL_VARIANTS_DATA = "clinical_variants";
+    public static final String SPLICE_DATA = "splice";
 
     public static final String CLINICAL_VARIANTS_FOLDER = "clinicalVariant";
     public static final String CLINVAR_XML_FILE = "ClinVarFullRelease_2020-02.xml.gz";