downloader: add cancer hotspot

opencb · Jan 2, 2024 · ce1767a · ce1767a
1 parent 89831a6
commit ce1767a
Show file tree

Hide file tree

Showing 3 changed files with 28 additions and 1 deletion.
diff --git a/cellbase-core/src/main/java/org/opencb/cellbase/core/config/DownloadProperties.java b/cellbase-core/src/main/java/org/opencb/cellbase/core/config/DownloadProperties.java
@@ -26,6 +26,7 @@ public class DownloadProperties {
     private EnsemblProperties ensembl;
     private EnsemblProperties ensemblGenomes;
     private URLProperties hgnc;
+    private URLProperties cancerHotspot;
     private URLProperties refSeq;
     private URLProperties refSeqFasta;
     private URLProperties refSeqProteinFasta;
@@ -517,6 +518,15 @@ public DownloadProperties setHgnc(URLProperties hgnc) {
         return this;
     }
 
+    public URLProperties getCancerHotspot() {
+        return cancerHotspot;
+    }
+
+    public DownloadProperties setCancerHotspot(URLProperties cancerHotspot) {
+        this.cancerHotspot = cancerHotspot;
+        return this;
+    }
+
     public static class EnsemblProperties {
 
         private DatabaseCredentials database;

diff --git a/cellbase-core/src/main/resources/configuration.yml b/cellbase-core/src/main/resources/configuration.yml
@@ -64,6 +64,9 @@ download:
   hgnc:
     host: https://ftp.ebi.ac.uk/pub/databases/genenames/hgnc/archive/monthly/tsv/hgnc_complete_set_2023-11-01.txt
     version: 2023-11-01
+  cancerHotspot:
+    host: https://www.cancerhotspots.org/files/hotspots_v2.xls
+    version: "v2"
   refSeq:
     host: https://ftp.ncbi.nih.gov/refseq/H_sapiens/annotation/GRCh38_latest/refseq_identifiers/GRCh38_latest_genomic.gtf.gz
   refSeqFasta:
@@ -161,6 +164,7 @@ download:
     host: https://old.dgidb.org/data/monthly_tsvs/2022-Feb/interactions.tsv
     version: "2022-02-01"
   cadd:
+    ## Nacho: Move to https://krishna.gs.washington.edu/download/CADD/v1.7-pre/GRCh38/whole_genome_SNVs.tsv.gz ASAP!
     host: https://krishna.gs.washington.edu/download/CADD/v1.6/GRCh38/whole_genome_SNVs.tsv.gz
   reactome:
     host: http://www.reactome.org/download/current/biopax.zip

diff --git a/cellbase-lib/src/main/java/org/opencb/cellbase/lib/download/GeneDownloadManager.java b/cellbase-lib/src/main/java/org/opencb/cellbase/lib/download/GeneDownloadManager.java
@@ -83,6 +83,7 @@ public List<DownloadFile> download() throws IOException, InterruptedException {
         downloadFiles.add(downloadMane(geneFolder));
         downloadFiles.add(downloadLrg(geneFolder));
         downloadFiles.add(downloadHgnc(geneFolder));
+        downloadFiles.add(downloadCancerHotspot(geneFolder));
         downloadFiles.add(downloadDrugData(geneFolder));
         downloadFiles.addAll(downloadGeneUniprotXref(geneFolder));
         downloadFiles.add(downloadGeneExpressionAtlas(geneFolder));
@@ -211,7 +212,7 @@ private DownloadFile downloadLrg(Path geneFolder) throws IOException, Interrupte
 
     private DownloadFile downloadHgnc(Path geneFolder) throws IOException, InterruptedException {
         if (speciesConfiguration.getScientificName().equals("Homo sapiens")) {
-            logger.info("Downloading LRG ...");
+            logger.info("Downloading HGNC ...");
             String url = configuration.getDownload().getHgnc().getHost();
             saveVersionData(EtlCommons.GENE_DATA, "HGNC_GENE", configuration.getDownload().getHgnc().getVersion(),
                     getTimeStamp(), Collections.singletonList(url), geneFolder.resolve("hgncVersion.json"));
@@ -221,6 +222,18 @@ private DownloadFile downloadHgnc(Path geneFolder) throws IOException, Interrupt
         return null;
     }
 
+    private DownloadFile downloadCancerHotspot(Path geneFolder) throws IOException, InterruptedException {
+        if (speciesConfiguration.getScientificName().equals("Homo sapiens")) {
+            logger.info("Downloading Cancer Hotspot ...");
+            String url = configuration.getDownload().getCancerHotspot().getHost();
+            saveVersionData(EtlCommons.GENE_DATA, "CANCER_HOTSPOT", configuration.getDownload().getHgnc().getVersion(),
+                    getTimeStamp(), Collections.singletonList(url), geneFolder.resolve("cancerHotspotVersion.json"));
+            String[] array = url.split("/");
+            return downloadFile(url, geneFolder.resolve(array[array.length - 1]).toString());
+        }
+        return null;
+    }
+
     private DownloadFile downloadGO(Path geneFolder) throws IOException, InterruptedException {
         if (speciesConfiguration.getScientificName().equals("Homo sapiens")) {
             logger.info("Downloading go annotation...");