test sra sqlite select

annot_biosample.txt

@@ -707,3 +707,4 @@ SAMN13675363		4822	Burkholderia pseudomallei			2018	Viet Nam		REMOVE: Not Yersin
 SAMN13698231		134	Burkholderia pseudomallei			1985	Thailand	Homo sapiens	REMOVE: Not Yersinia pestis
 SAMN13698233		134K	Burkholderia pseudomallei			2019	Thailand		REMOVE: Not Yersinia pestis
 SAMN08892357		139	Yersinia pestis			1966	Kazakhstan	Rhombomis opinus	REMOVE: Pairwise alignment 60 percent coverage of reference 
+SAMEA3541827	PRJEB10885	RISE509	Yersinia pestis	ERS848976	SAMEA3325401	"14C cal BC low: -2677, 14C cal BC High: -2887"	Russia: Bateni	Homo sapiens	KEEP

nextflow.config

@@ -104,9 +104,9 @@ params{
   iqtree_rng = "4154541355"
 
   // SQLite q
-  // sqlite_select_command = "\'SELECT AssemblyFTPGenbank FROM Master WHERE BioSampleComment NOT LIKE \"%REMOVE%\"\'"
-  sqlite_select_command = "\"SELECT AssemblyFTPGenbank FROM Master WHERE (BioSampleComment NOT LIKE \'%REMOVE%\') AND (TRIM(BioSampleHost) > \'\') AND (TRIM(LOWER(BioSampleHost)) IS NOT \'missing\') AND (TRIM(BioSampleCollectionDate) > \'\') AND (TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'missing\' AND TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'not applicable\' AND TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'unknown\' AND TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'n/a\' ) AND (TRIM(BioSampleGeographicLocation) > '') AND (TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'missing\') AND (TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'missing\') AND (TRIM(AssemblyFTPGenbank) > \'\') \""
-
+  // sqlite_select_command_asm = "\'SELECT AssemblyFTPGenbank FROM Master WHERE BioSampleComment NOT LIKE \"%REMOVE%\"\'"
+  sqlite_select_command_asm = "\"SELECT AssemblyFTPGenbank FROM Master WHERE (BioSampleComment NOT LIKE \'%REMOVE%\') AND (TRIM(BioSampleHost) > \'\') AND (TRIM(LOWER(BioSampleHost)) IS NOT \'missing\') AND (TRIM(BioSampleCollectionDate) > \'\') AND (TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'missing\' AND TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'not applicable\' AND TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'unknown\' AND TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'n/a\' ) AND (TRIM(BioSampleGeographicLocation) > '') AND (TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'missing\') AND (TRIM(LOWER(BioSampleCollectionDate)) IS NOT \'missing\') AND (TRIM(AssemblyFTPGenbank) > \'\') \""
+  sqlite_select_command_sra = "\"SELECT SRARunAccession FROM Master WHERE (BioSampleComment LIKE \'%KEEP%\')\""
   // MultiQC
   multiqc_config = "$baseDir/multiqc_config.yaml"
 

pipeline.nf

@@ -241,6 +241,7 @@ if( (params.sqlite || ( params.ncbimeta_update && params.ncbimeta_annot) ) && !p
     */
     // Other variables and config
     tag "$sqlite"
+    echo true
     publishDir "${outdir}/sqlite_import", mode: 'copy'
     // Set the sqlite channel to update or sqlite import depending on ncbimeta mode
     // TO DO: catch if both parameters are specified!!!
@@ -261,14 +262,20 @@ if( (params.sqlite || ( params.ncbimeta_update && params.ncbimeta_annot) ) && !p
     // Shell script to execute
     script:
     """
-    sqlite3 ${sqlite} ${params.sqlite_select_command} | grep . | head -n ${params.max_datasets} | sed -E -e 's/ |;/\\n/g' | while read line;
+    # Select the Genbank Assemblies
+    sqlite3 ${sqlite} ${params.sqlite_select_command_asm} | grep . | head -n ${params.max_datasets} | sed -E -e 's/ |;/\\n/g' | while read line;
     do
       if [[ ! -z \$line ]]; then
         asm_ftp=`echo \$line | \
             awk -F "/" -v suffix=${params.genbank_assembly_gz_suffix} '{print \$0 FS \$NF suffix}'`;
         echo \$asm_ftp >> ${params.file_assembly_for_download_ftp}
       fi;
     done;
+    # Select the SRA Run Accessions
+    sqlite3 ${sqlite} ${params.sqlite_select_command_sra} | grep . | head -n ${params.max_datasets} | sed -E -e 's/ |;/\\n/g' | while read line;
+    do
+      echo "\$line";
+    done;
     """
   }
 
@@ -316,6 +323,39 @@ if (!params.skip_assembly_download){
   }
 
 }
+
+// -------------------------------------------------------------------------- //
+//                                SRA Download                                //
+// -------------------------------------------------------------------------- //
+/*
+process sra_download{
+
+  Input:
+  ch_():
+
+  Output:
+  ch_ ():
+
+  Publish:
+
+  // Other variables and config
+  tag ""
+  publishDir
+
+  // IO and conditional behavior
+  input:
+
+  output:
+
+
+  // Shell script to execute
+  script:
+  """
+  """
+}
+*/
+
+
 // -------------------------------------------------------------------------- //
 //                           Reference Genome Processing                      //
 // -------------------------------------------------------------------------- //