updated GDB config files

config/GDB/config.yaml

@@ -3,7 +3,7 @@
 
 # Steps to be done
 # steps: ["preprocessing", "assembly", "mapping", "annotation", "analysis", "taxonomy"]
-steps: ["assembly", "mapping", "annotation", "analysis", "taxonomy"]
+steps: ["preprocessing", "assembly", "mapping", "annotation", "analysis", "taxonomy"]
 steps_annotation: ["diamond", "rgi", "plasflow", "minced", "barrnap"] # prodigal is run in any case
 steps_analysis: ["quast", "cdhit", "mash_dist"]
 steps_taxonomy: ["kraken2", "kaiju"]
@@ -72,6 +72,17 @@ fastp:
     threads: 10
     min_length: 40
 
+# rRNA gene filtering
+sortmerna:
+    threads: 20
+    # References to be used (w/ md5sums)
+    refs: [
+        # db6e72022cf650c4b33bd888b92a0391  silva-bac-16s-id90.fasta
+        "/mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-bac-16s-id90.fasta",
+        # f347d2f8f8ffbfa28c785e3a9fe3db79  silva-bac-23s-id98.fasta
+        "/mnt/irisgpfs/projects/ecosystem_biology/local_tools/IMP3/databases/sortmerna/silva-bac-23s-id98.fasta"
+    ]
+
 # FastQ QC
 # https://www.bioinformatics.babraham.ac.uk/projects/fastqc/
 fastqc:
@@ -83,12 +94,10 @@ fastqc:
 
 # List of assemblers for different read types: assembler names MUST be UNIQUE
 assemblers:
-    # sr: ["megahit", "metaspades", "megahitmetatracon", "metaspadesmetatracon"]
-    # lr: ["flye"]
-    # hy: ["metaspadeshybrid", "operams", "metaspadeshybridmetatracon", "operamsmetatracon"]
     sr: ["megahit", "metaspades"]
     lr: ["flye", "wtdbg2", "canu"]
     hy: ["metaspadeshybrid", "operams"]
+    hyhy: []
 
 # https://github.com/fenderglass/Flye
 flye:

config/GDB/sbatch.sh

@@ -12,7 +12,7 @@
 # conda env name or path
 ONTP_ENV="ONT_pilot"
 # number of concurrent jobs
-ONTP_JOBS=5
+ONTP_JOBS=10
 # config files
 ONTP_CONFIG="config/GDB/config.yaml"
 ONTP_SLURM="config/GDB/slurm.yaml"

config/GDB/slurm.yaml

@@ -27,6 +27,14 @@ fastp_sr:
     n: 1
     explicit: ""
 
+sortmerna_filt:
+    time: "00-12:00:00"
+    partition: "batch"
+    qos: "qos-batch"
+    nodes: 1
+    n: 1
+    explicit: ""
+
 # Assembly
 assembly_lr_flye:
     time: "00-8:00:00"
@@ -143,7 +151,7 @@ polishing_lr_racon:
     explicit: ""
 
 polishing_lr_medaka:
-    time: "00-12:00:00"
+    time: "01-00:00:00"
     partition: "bigmem"
     qos: "qos-bigmem"
     nodes: 1