Add option to download VFDB in db_setup workflow.

bin/zdb

@@ -119,6 +119,10 @@ elif [[ "$1" == "setup" ]]; then
 				db_setup_args="${db_setup_args} --pfam"
 				shift
 				;;
+			--vfdb)
+				db_setup_args="${db_setup_args} --vfdb"
+				shift
+				;;
 			--singularity_dir=*)
 				singularity_dir=${i#*=}
 				shift
@@ -135,6 +139,7 @@ elif [[ "$1" == "setup" ]]; then
 				echo " --ko: downloads and setups the hmm profiles of the ko database"
 				echo " --pfam: downloads and setups up the hmm profiles of the PFAM protein domains"
 				echo " --swissprot: downloads and indexes the swissprot database "
+				echo " --vfdb: downloads the virulence factor database "
 				echo ""
 				echo "Other parameters:"
 				echo " --dir: directory where to store the reference databases (defaults zdb_ref in the current directory)"

db_setup.nf

@@ -149,6 +149,24 @@ process prepare_swissprot {
     """
 }
 
+process download_vfdb {
+
+    publishDir "$params.vf_db", mode: "move"
+
+    output:
+        tuple path("vfdb.fasta"), path("VFs.xls")
+
+    script:
+    """
+    wget http://www.mgc.ac.cn/VFs/Down/VFDB_setB_nt.fas.gz
+    wget http://www.mgc.ac.cn/VFs/Down/VFs.xls.gz
+    gunzip < VFDB_setB_nt.fas.gz > vfdb.fasta
+    gunzip < VFs.xls.gz > VFs.xls
+    rm VFDB_setB_nt.fas.gz
+    rm VFs.xls.gz
+    """
+}
+
 workflow setup_cogg_db {
     download_cog_cdd() | setup_cog_cdd
 }
@@ -174,6 +192,10 @@ workflow setup_swissprot_db {
     download_swissprot() | prepare_swissprot
 }
 
+workflow setup_vfdb {
+    download_vfdb()
+}
+
 workflow {
     if( params.cog )
         setup_cogg_db()
@@ -189,4 +211,7 @@ workflow {
 
     if ( params.blast_swissprot )
         setup_swissprot_db()
+
+    if ( params.vfdb )
+        setup_vfdb()
 }

nextflow.config

@@ -11,6 +11,7 @@ params.pfam_db = "${params.base_db}/pfam/"
 params.cog_db = "${params.base_db}/cog/"
 params.ko_db = "${params.base_db}/kegg/"
 params.refseq_db = "${params.base_db}/refseq/"
+params.vf_db = "${params.base_db}/vfdb/"
 
 params.results_dir = "zdb/results"
 
@@ -22,6 +23,7 @@ params.cog = false
 params.ko = false
 params.pfam = false
 params.amr = false
+params.vfdb = false
 
 params.checkm_args = "domain Bacteria"
 

testing/pipelines/test_db_setup.py

@@ -108,3 +108,20 @@ def test_creating_swissprot_db(self):
         self.assertItemsEqual(
             expected_files,
             os.listdir(os.path.join(self.ref_db_dir, "uniprot", "swissprot")))
+
+    def test_creating_vf_db(self):
+        self.nf_params["vfdb"] = "true"
+        execution = self.execute_pipeline()
+        self.assert_success(execution)
+
+        self.assertEqual([proc.name for proc in execution.process_executions],
+                         ["setup_vfdb:download_vfdb"])
+
+        download_process = execution.process_executions[0]
+        # Files are moved to zdb_ref/vfdb
+        expected_files = ['vfdb.fasta',
+                          'VFs.xls']
+        self.assert_created_files(download_process, [])
+        self.assertItemsEqual(
+            expected_files,
+            os.listdir(os.path.join(self.ref_db_dir, "vfdb")))