Update kraken db (#30)

* fix: update to new kraken-plusfp database version

* fix: update kraken memory for larger index

* fix: update kraken2 db download feature

* fix: update new bacdb name

* fix: update build targets for new kraken bac db

Author: rroutsong

config/templates/tools.json

@@ -32,7 +32,7 @@
                 "FASTAWITHADAPTERSETD": "resources/TruSeq_and_nextera_adapters.consolidated.fa",
                 "FASTQ_SCREEN_CONFIG": "resources/fastq_screen.conf",
                 "FASTQ_SCREEN_CONFIG2": "resources/fastq_screen_2.conf",
-                "KRAKENBACDB": "/fdb/kraken/20180907_standard_kraken2",
+                "KRAKENBACDB": "/data/OpenOmics/references/kraken/k2_pluspf_20241228",
                 "Trimsettings": "---------------------------- TRIMMOMATIC parameters ----------------------------------------------------------",
                 "LEADINGQUALITY": 10,
                 "TRAILINGQUALITY": 10,
@@ -61,4 +61,4 @@
             "WIGTYPE": "None"
         }
     }
-}
+}

resources/cluster.json

@@ -32,12 +32,12 @@
 	"kraken_pe": {
 		"cpus-per-task": "24",
 		"gres": "lscratch:256",
-		"mem": "110g"
+		"mem": "240g"
 	},
 	"kraken_se": {
 		"cpus-per-task": "24",
 		"gres": "lscratch:256",
-		"mem": "110g"
+		"mem": "240g"
 	},
 	"pca": {
 		"mem": "88g",

rna-seek

@@ -652,7 +652,7 @@ def setup(sub_args, ifiles, repo_path, output_path):
         # Update paths to shared resources directory
         config['bin']['rnaseq']['tool_parameters']['FASTQ_SCREEN_CONFIG'] = os.path.join(sub_args.shared_resources, "fastq_screen_db", "fastq_screen_p1.conf")
         config['bin']['rnaseq']['tool_parameters']['FASTQ_SCREEN_CONFIG2'] = os.path.join(sub_args.shared_resources, "fastq_screen_db", "fastq_screen_p2.conf")
-        config['bin']['rnaseq']['tool_parameters']['KRAKENBACDB'] = os.path.join(sub_args.shared_resources, "20180907_standard_kraken2")
+        config['bin']['rnaseq']['tool_parameters']['KRAKENBACDB'] = os.path.join(sub_args.shared_resources, "k2_pluspf_20241228")
 
     # Save config to output directory
     print("\nGenerating config file in '{}'... ".format(os.path.join(output_path, 'config.json')), end = "")

workflow/rules/build.smk

@@ -134,7 +134,7 @@ rule all:
         provided(expand(join(SHARED_PATH, "fastq_screen_db", "fastq_screen_p{ext}.conf"), ext=["1", "2"]), SHARED_PATH),
         # Kraken2 Database,
         # conditional runs with --shared-resources option
-        provided(expand(join(SHARED_PATH, "20180907_standard_kraken2", "{ref}.k2d"), ref=["hash", "opts", "taxo"]), SHARED_PATH),
+        provided(expand(join(SHARED_PATH, "k2_pluspf_20241228", "{ref}.k2d"), ref=["hash", "opts", "taxo"]), SHARED_PATH),
 
 
 rule rsem:
@@ -640,16 +640,16 @@ rule kraken2_db:
         fa=REFFA,
         gtf=GTFFILE,
     output:
-        expand(join(SHARED_PATH, "20180907_standard_kraken2", "{ref}.k2d"), ref=["hash", "opts", "taxo"]),
+        expand(join(SHARED_PATH, "k2_pluspf_20241228", "{ref}.k2d"), ref=["hash", "opts", "taxo"]),
     params:
         rname='bl:kraken2_db',
-        outfh=join(SHARED_PATH, "20180907_standard_kraken2.tar"),
-        tarfile="20180907_standard_kraken2.tar",
+        outfh=join(SHARED_PATH, "k2_pluspf_20241228.tar.gz"),
+        tarfile="k2_pluspf_20241228.tar.gz",
         outdir=SHARED_PATH,
     container: config['images']['build_rnaseq']
     shell: """
     wget https://hpc.nih.gov/~OpenOmics/common/{params.tarfile} -O {params.outfh}
-    tar vxf {params.outfh} -C {params.outdir} && rm {params.outfh}
+    tar zvxf {params.outfh} -C {params.outdir} && rm {params.outfh}
     """