updated GDB config files (issue #54)

config/GDB/config.yaml

@@ -3,7 +3,7 @@
 
 # Steps to be done
 # steps: ["preprocessing", "assembly", "mapping", "annotation", "analysis", "taxonomy"]
-steps: ["preprocessing", "assembly", "mapping", "annotation", "analysis", "taxonomy"]
+steps: ["preprocessing"]
 steps_annotation: ["diamond", "rgi", "plasflow", "minced", "barrnap"] # prodigal is run in any case
 steps_analysis: ["quast", "cdhit", "mash_dist"]
 steps_taxonomy: ["kraken2", "kaiju"]
@@ -16,7 +16,7 @@ work_dir: "/scratch/users/vgalata/ont_pilot"
 
 # Paths WITHIN the working directory
 # directory containing required DBs (should be writeable)
-db_dir: "dbs"
+db_dir: "/mnt/lscratch/users/vgalata/ONT_pilot_DBs"
 # results directory (will be created in work_dir)
 results_dir: "results"
 
@@ -161,7 +161,7 @@ samtools:
 # https://github.com/bbuchfink/diamond
 diamond:
     threads: 20
-    db: "/work/projects/ecosystem_biology/local_tools/databases/nr_uniprot_trembl.dmnd" # TODO: data download
+    db: "nr_uniprot_trembl.dmnd" # file name in "dbs" folder
 
 # CRISPR
 # https://github.com/dnasko/CASC
@@ -202,25 +202,19 @@ cdhit:
 # https://github.com/BioInfoTools/BBMap/
 bbmap:
     threads: 10
-    # References to be used (w/ md5sums)
-    rrna_refs: [
-        # c0cd2aa2e84e3e3977859c34feb63cd5  /mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/rfam-5.8s-database-id98.fasta
-        # 703e4c270ab0a578deb4800c33b36367  /mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/rfam-5s-database-id98.fasta
-        # 8b4e6c6f17f6f35444a60fdc915e052c  /mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-arc-16s-id95.fasta
-        # ca4edcdddb98d7868f93e2308e297704  /mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-arc-23s-id98.fasta
-        # db6e72022cf650c4b33bd888b92a0391  /mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-bac-16s-id90.fasta
-        # f347d2f8f8ffbfa28c785e3a9fe3db79  /mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-bac-23s-id98.fasta
-        # 878a413765d09c3ec75409fb1d1573f1  /mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-euk-18s-id95.fasta
-        # cbb973e63f52981bd591de0404df5839  /mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-euk-28s-id98.fast
-        "/mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/rfam-5.8s-database-id98.fasta",
-        "/mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/rfam-5s-database-id98.fasta",
-        "/mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-arc-16s-id95.fasta",
-        "/mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-arc-23s-id98.fasta",
-        "/mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-bac-16s-id90.fasta",
-        "/mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-bac-23s-id98.fasta",
-        "/mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-euk-18s-id95.fasta",
-        "/mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-euk-28s-id98.fasta"
+    rrna_refs: [ # file names in "dbs" folder
+        "sortmerna/rfam-5.8s-database-id98.fasta",
+        "sortmerna/rfam-5s-database-id98.fasta",
+        "sortmerna/silva-arc-16s-id95.fasta",
+        "sortmerna/silva-arc-23s-id98.fasta",
+        "sortmerna/silva-bac-16s-id90.fasta",
+        "sortmerna/silva-bac-23s-id98.fasta",
+        "sortmerna/silva-euk-18s-id95.fasta",
+        "sortmerna/silva-euk-28s-id98.fasta"
     ]
+    host_refs: # keep empty if no host specific reads should be removed
+        # key: url of GZ archive
+        GCF_000001405.38_GRCh38.p12: "ftp.ncbi.nlm.nih.gov/genomes/all/GCF/000/001/405/GCF_000001405.38_GRCh38.p12/GCF_000001405.38_GRCh38.p12_genomic.fna.gz"
 
 # Assembly quality
 # https://github.com/ablab/quast
@@ -246,8 +240,8 @@ quast:
 # https://github.com/DerrickWood/kraken2
 kraken2:
     threads: 10
-    db:
-        maxikraken: "/scratch/users/bkunath/Kraken2/maxikraken2_1903_140GB/"
+    db: # dir. name in "dbs" folder
+        maxikraken: "maxikraken2_1903_140GB"
     class:
         sr: "--gzip-compressed --paired"
         lr: ""
@@ -258,13 +252,14 @@ kraken2:
 # https://github.com/bioinformatics-centre/kaiju
 kaiju:
     threads: 10
-    db: # key = basename of *.fmi
-        kaiju_db_nr_euk: "/mnt/isilon/projects/ecosystem_biology/databases/kaiju/kaiju_db_nr_euk_2020-05-25"
+    db: # dir. name in "dbs" folder
+        # key = basename of *.fmi
+        kaiju_db_nr_euk: "kaiju_db_nr_euk_2020-05-25"
     ranks: ["phylum", "class", "order", "family", "genus", "species"]
 
-# # XXX
-# GTDBTK:
-#     DATA: "/home/users/sbusi/apps/db/gtdbtk/release89"
+# XXX
+GTDBTK: # dir. name in "dbs" folder
+    DATA: "gtdbtk_release89"
 
 ##############################
 # MISC

config/GDB/slurm.yaml

@@ -35,6 +35,30 @@ rm_rrna_bbmap:
     n: 1
     explicit: ""
 
+rm_host_bbmap_sr_metat:
+    time: "00-4:00:00"
+    partition: "bigmem"
+    qos: "qos-bigmem"
+    nodes: 1
+    n: 1
+    explicit: ""
+
+rm_host_bbmap_sr_metag:
+    time: "00-4:00:00"
+    partition: "bigmem"
+    qos: "qos-bigmem"
+    nodes: 1
+    n: 1
+    explicit: ""
+
+rm_host_bbmap_lr_metag:
+    time: "00-4:00:00"
+    partition: "bigmem"
+    qos: "qos-bigmem"
+    nodes: 1
+    n: 1
+    explicit: ""
+
 # Assembly
 assembly_lr_flye:
     time: "00-8:00:00"