lib: update configuration file to download last versions of ClinVar a…

…nd Gwas, files to download; and update Cosmic version, #TASK-6347 On branch TASK-6347 Changes to be committed: modified: cellbase-core/src/main/resources/configuration.yml modified: cellbase-lib/src/main/java/org/opencb/cellbase/lib/builders/clinical/variant/CosmicIndexer.java modified: cellbase-lib/src/main/java/org/opencb/cellbase/lib/download/ClinicalDownloadManager.java
opencb · Jun 4, 2024 · 499407d · 499407d
1 parent 9ab2fe6
commit 499407d
Show file tree

Hide file tree

Showing 3 changed files with 12 additions and 11 deletions.
diff --git a/cellbase-core/src/main/resources/configuration.yml b/cellbase-core/src/main/resources/configuration.yml
@@ -106,11 +106,15 @@ download:
   clinvar:
 #    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/ClinVarFullRelease_2021-07.xml.gz
 #    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/ClinVarFullRelease_2022-02.xml.gz
-    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/ClinVarFullRelease_2022-11.xml.gz
+#    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/ClinVarFullRelease_2022-11.xml.gz
+    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/RCV_xml_old_format/ClinVarFullRelease_2024-05.xml.gz
+    version: 2024-05
   clinvarVariation:
 #    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/clinvar_variation/ClinVarVariationRelease_2021-07.xml.gz
 #    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/clinvar_variation/ClinVarVariationRelease_2022-02.xml.gz
-    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/clinvar_variation/ClinVarVariationRelease_2022-11.xml.gz
+#    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/clinvar_variation/ClinVarVariationRelease_2022-11.xml.gz
+    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/VCV_xml_old_format/ClinVarVariationRelease_2024-05.xml.gz
+    version: 2024-05
   clinvarSummary:
     host: http://ftp.ncbi.nlm.nih.gov/pub/clinvar/tab_delimited/variant_summary.txt.gz
   clinvarVariationAllele:

diff --git a/...se-lib/src/main/java/org/opencb/cellbase/lib/builders/clinical/variant/CosmicIndexer.java b/...se-lib/src/main/java/org/opencb/cellbase/lib/builders/clinical/variant/CosmicIndexer.java
@@ -41,7 +41,7 @@ public class CosmicIndexer extends ClinicalIndexer {
     private Pattern mutationGRCh37GenomePositionPattern;
     private Pattern snvPattern;
 
-    private static final String COSMIC_VERSION = "v95";
+    private static final String COSMIC_VERSION = "v99";
 
     private static final int GENE_NAMES_COLUMN = 0;
     private static final int HGNC_COLUMN = 3;

diff --git a/cellbase-lib/src/main/java/org/opencb/cellbase/lib/download/ClinicalDownloadManager.java b/cellbase-lib/src/main/java/org/opencb/cellbase/lib/download/ClinicalDownloadManager.java
@@ -81,8 +81,10 @@ public List<DownloadFile> downloadClinical() throws IOException, InterruptedExce
             url = configuration.getDownload().getClinvarVariationAllele().getHost();
             downloadFiles.add(downloadFile(url, clinicalFolder.resolve(EtlCommons.CLINVAR_VARIATION_ALLELE_FILE).toString()));
             clinvarUrls.add(url);
-            saveVersionData(EtlCommons.CLINICAL_VARIANTS_DATA, CLINVAR_NAME, getClinVarVersion(), getTimeStamp(), clinvarUrls,
-                    clinicalFolder.resolve("clinvarVersion.json"));
+            saveVersionData(EtlCommons.CLINICAL_VARIANTS_DATA, CLINVAR_NAME, configuration.getDownload().getClinvar()
+                            .getVersion(), getTimeStamp(), clinvarUrls, clinicalFolder.resolve("clinvarVersion.json"));
+
+            logger.info("\t\tDone");
 
             // Gwas catalog
             logger.info("\t\tDownloading GWAS catalog file ...");
@@ -91,6 +93,7 @@ public List<DownloadFile> downloadClinical() throws IOException, InterruptedExce
             downloadFiles.add(downloadFile(url, clinicalFolder.resolve(EtlCommons.GWAS_FILE).toString()));
             saveVersionData(EtlCommons.CLINICAL_VARIANTS_DATA, GWAS_NAME, gwasCatalog.getVersion(), getTimeStamp(),
                     Collections.singletonList(url), clinicalFolder.resolve("gwasVersion.json"));
+            logger.info("\t\tDone");
 
 //            List<String> hgvsList = getDocmHgvsList();
 //            if (!hgvsList.isEmpty()) {
@@ -236,10 +239,4 @@ private List<String> getDocmHgvsList() throws IOException {
 
         return hgvsList;
     }
-
-    private String getClinVarVersion() {
-        // ftp://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/ClinVarFullRelease_2015-12.xml.gz
-        return configuration.getDownload().getClinvar().getHost().split("_")[1].split("\\.")[0];
-    }
-
 }