lib: update GeneDownloadManager to call the script gene_extra_info.pl…

…, #TASK-5575, #TASK-5564

cellbase-app/app/scripts/ensembl-scripts/ensembl_canonical.pl

@@ -36,13 +36,26 @@
 
 $query->formatter("TSV");
 
-open (ENSEMBL_CANONICAL, ">$outdir/ensembl_canonical.txt") || die "Cannot open ensembl_canonical.txt file";
+# Open the file for writing
+open(my $fh, '>', "$outdir/ensembl_canonical.txt") or die "Cannot open ensembl_canonical.txt file: $!";
+
+# Save the original stdout
+my $original_stdout = *STDOUT;
+open(STDOUT, '>&', $fh) or die "Can't redirect STDOUT: $!";
 
 my $query_runner = BioMart::QueryRunner->new();
 
 # to obtain unique rows only
 $query_runner->uniqueRowsOnly(1);
 $query_runner->execute($query);
 #$query_runner->printHeader();
-print ENSEMBL_CANONICAL $query_runner->printResults();
+#print ENSEMBL_CANONICAL $query_runner->printResults();
+# Call printResults which prints to STDOUT (now redirected to the file)
+$query_runner->printResults();
 #$query_runner->printFooter();
+
+# Restore the original stdout
+open(STDOUT, '>&', $original_stdout) or die "Can't restore STDOUT: $!";
+
+# Close the filehandle
+close($fh) or die "Failed to close file: $!";

cellbase-app/app/scripts/ensembl-scripts/genome_info.pl

@@ -143,7 +143,7 @@
 
 sub print_parameters {
     print "Parameters: ";
-    print "species: $species, outfile: $outfile, ";
+    print "species: $species, assembly: $assembly, outfile: $outfile, ";
     print "ensembl-registry: $ENSEMBL_REGISTRY, ";
     print "ensembl-host: $ENSEMBL_HOST, ensembl-port: $ENSEMBL_PORT, ";
     print "ensembl-user: $ENSEMBL_USER, verbose: $verbose, help: $help";

cellbase-lib/src/main/java/org/opencb/cellbase/lib/EtlCommons.java

@@ -110,6 +110,8 @@ public final class EtlCommons {
     public static final String REFSEQ_RNA_FNA_FILE_ID = "RNA_FNA";
 
     // Gene annotation
+    public static final String ENSEMBL_CANONICAL_DATA = "ensembl_canonical";
+    public static final String GENE_EXTRA_INFO_DATA = "gene_extra_info";
     //   - MANE Select
     public static final String MANE_SELECT_DATA = "MANE Select";
     // Must match the configuration file
@@ -354,6 +356,8 @@ public final class EtlCommons {
         dataNamesMap.put(GENOME_DATA, "Genome");
         dataNamesMap.put(GENOME_INFO_DATA, "Genome Info");
         dataNamesMap.put(GENE_DATA, "Gene");
+        dataNamesMap.put(ENSEMBL_CANONICAL_DATA, "Ensembl canonical");
+        dataNamesMap.put(GENE_EXTRA_INFO_DATA, "Gene extra info");
         dataNamesMap.put(GENE_ANNOTATION_DATA, "Gene Annotation");
         dataNamesMap.put(MANE_SELECT_DATA, "MANE Select");
         dataNamesMap.put(LRG_DATA, "LRG");

cellbase-lib/src/main/java/org/opencb/cellbase/lib/download/GeneDownloadManager.java

@@ -58,7 +58,10 @@ public List<DownloadFile> download() throws IOException, InterruptedException, C
         downloadFiles.addAll(downloadEnsemblData(ensemblDownloadPath));
 
         // Ensembl canonical
-        downloadEnsemblCanonical();
+        downloadEnsemblCanonical(geneDownloadPath);
+
+        // Gene extra info
+        downloadGeneExtraInfo(geneDownloadPath);
 
         // RefSeq
         downloadFiles.addAll(downloadRefSeq(refSeqDownloadPath));
@@ -151,20 +154,43 @@ private List<DownloadFile> downloadRefSeq(Path refSeqDownloadPath) throws IOExce
         return downloadFiles;
     }
 
-    public void downloadEnsemblCanonical() throws IOException, CellBaseException {
-        logger.info(DOWNLOADING_LOG_MESSAGE, getDataName(GENOME_INFO_DATA));
-        Path sequenceFolder = downloadFolder.resolve(GENOME_DATA);
-        Files.createDirectories(sequenceFolder);
+    public void downloadEnsemblCanonical(Path geneDownloadPath) throws IOException, CellBaseException {
+        logger.info(DOWNLOADING_LOG_MESSAGE, getDataName(ENSEMBL_CANONICAL_DATA));
 
         String dockerImage = "opencb/cellbase-builder:" + GitRepositoryState.get().getBuildVersion();
         try {
             // Build command line to run Perl script via docker image
             // Output binding
             AbstractMap.SimpleEntry<String, String> outputBinding = new AbstractMap.SimpleEntry<>(
-                    sequenceFolder.toAbsolutePath().toString(), "/tmp");
+                    geneDownloadPath.toAbsolutePath().toString(), "/tmp");
 
             // Params
             String params = "/opt/cellbase/scripts/ensembl-scripts/ensembl_canonical.pl"
+                    + " --species \"" + speciesConfiguration.getId() + "\""
+                    + " --assembly \"" + assemblyConfiguration.getName() + "\""
+                    + " --outdir \"" + outputBinding.getValue() + "\"";
+
+            // Execute perl script in docker
+            DockerUtils.run(dockerImage, null, outputBinding, params, null);
+        } catch (Exception e) {
+            throw new CellBaseException("Error executing Perl script from Docker " + dockerImage, e);
+        }
+
+        logger.info(DOWNLOADING_DONE_LOG_MESSAGE, getDataName(ENSEMBL_CANONICAL_DATA));
+    }
+
+    public void downloadGeneExtraInfo(Path geneDownloadPath) throws IOException, CellBaseException {
+        logger.info(DOWNLOADING_LOG_MESSAGE, getDataName(GENE_EXTRA_INFO_DATA));
+
+        String dockerImage = "opencb/cellbase-builder:" + GitRepositoryState.get().getBuildVersion();
+        try {
+            // Build command line to run Perl script via docker image
+            // Output binding
+            AbstractMap.SimpleEntry<String, String> outputBinding = new AbstractMap.SimpleEntry<>(
+                    geneDownloadPath.toAbsolutePath().toString(), "/tmp");
+
+            // Params
+            String params = "/opt/cellbase/scripts/ensembl-scripts/gene_extra_info.pl"
                     + " --species \"" + speciesConfiguration.getId() + "\""
                     + " --outdir \"" + outputBinding.getValue() + "\"";
 
@@ -174,7 +200,7 @@ public void downloadEnsemblCanonical() throws IOException, CellBaseException {
             throw new CellBaseException("Error executing Perl script from Docker " + dockerImage, e);
         }
 
-        logger.info(DOWNLOADING_DONE_LOG_MESSAGE, getDataName(GENOME_INFO_DATA));
+        logger.info(DOWNLOADING_DONE_LOG_MESSAGE, getDataName(GENE_EXTRA_INFO_DATA));
     }
 
     private DownloadFile downloadMane(Path geneDownloadPath) throws IOException, InterruptedException, CellBaseException {